Loading page content
Preparing content…
SOC 15-1221 · O*NET 15-1221.00
Conduct basic computer and information science research as a theorist, designer or inventor, and develop solutions to problems in the computer hardware and software field: analyze problems to develop solutions involving computer hardware and software; apply theoretical expertise and innovation to create or apply new technology; conduct logical analysis of business and scientific and engineering and technical problems and build mathematical models for computer solution; evaluate project plans and proposals to determine feasibility; design computers and the software that runs on them; maintain network hardware and software and direct security measures and monitor availability; participate in interdisciplinary projects such as virtual reality and human-computer interaction or robotics; assign tasks and set goals and performance standards and evaluate work; direct the daily operations of a department and coordinate projects with other departments; participate in staffing decisions and supervise the training of subordinates; and approve, prepare, monitor and adjust operating budgets.
Model impact index for three equally weighted representative core tasks; not a measured share of working time or a job-loss probability. floor((50+50+75)/3+0.5)=58; fermatmind-ai-impact-v1; 2026-10-18.
In one sentence: this is an occupation in which "whether the question is asked correctly, whether the method holds up to reproduction, and whether the results are written clearly" determines success more than programming skill; if you are willing to treat reproducibility and honest boundaries as a method rather than a procedural requirement, it is worth serious consideration.
Education, on-the-job training, professional certifications, licenses and job authorization must be verified separately: O*NET lists this occupation as Job Zone Five (extensive preparation), and the page's education distribution is expressed in three tiers: respondents said "some positions require a bachelor's degree," "some positions require a doctorate" and "some positions require a master's degree" (the page uses these three tiers and gives no percentages); the page also lists 16 national certifications (such as Certified Computer Scientist, Certified Function Point Practitioner／Specialist, Certified LabWindows/CVI Developer, Certified in Open Source Intelligence, among others), but a certification is not equal to a degree, an employer appointment, signing authority or research qualifications; research ethics review involving human participants, data and code management, and intellectual property and technology export must be confirmed item by item according to the applicable jurisdiction and institution and funder requirements; this page describes tasks and boundaries based on O*NET and public sources, and does not present education or certifications as statutory qualifications.
O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. The editorial process runs from problem, prototype and experiment records to report and exchange and iteration; not all work must undergo peer review first. Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority.
Career direction comparison
The table below compares the positions and occupations adjacent to this occupation: computer and information systems managers, architectural and engineering managers, natural sciences managers, computer systems analysts, database architects and data scientists. The four columns each answer the same kind of question: what the direction does, where the key difference from this occupation lies, and what kind of person is better suited to choosing it.
| Career direction | Core work and output | Key differences | Best-fit choice |
|---|---|---|---|
| Computer and information systems managers | Mainly department management and enterprise system delivery and operations records. | Computer and information systems managers coordinate the people, budget, projects and compliance of the information technology department, aimed at the stable operation and delivery of business systems; this occupation centers on basic research and the creation and verification of new principles and new technology. | People willing to coordinate the IT department and project delivery suit that direction; people willing to do basic research and verify new methods suit this occupation. |
AI impact on this career
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred. AI Impact Score 58% is an editorial estimate. Model impact index for three equally weighted representative core tasks; not a measured share of working time or a job-loss probability. floor((50+50+75)/3+0.5)=58; fermatmind-ai-impact-v1; 2026-10-18.The three tasks, ratings and sources are listed in the FermatMind method source record on this page.
FermatMind AI Impact Score estimation method v1
FermatMind / 费马测试 · 2026-10-18
Chinese mainland salary reference
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority.
U.S. salary reference (query entry and scope)
OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered. BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
| Wage statistical position | Annual wage reference | Conversion boundaries | How to understand this position |
|---|---|---|---|
| The 10th percentile | 82,200 USD/year (2025 10th percentile) | Statistical percentiles, not personal monthly income or an offer | The 25th percentile is 103,570 USD/year; a statistical position, not starting pay. |
Career fit guide
Direct answer: the signals that best match this occupation's tasks are "willingness to write the problem to a testable degree and make results reproducible" and "ability to keep advancing through long-term uncertainty and failed experiments"; the interest code is IC, but the code only describes occupational-level interest characteristics and does not determine whether an individual is suited.
You may be better suited to： People willing to write the problem to a testable degree and make results reproducible, able to keep advancing through long-term uncertainty and failed experiments, and willing to record the method and experiments and limitations for review.
These situations call for caution： People who find it hard to accept having their methods and results challenged, who tend to adjust metrics or selectively present results to beautify outcomes, or who do not take data and code and disclosure and ethics requirements seriously enough.
Do an entry exercise once： Use the fictitious materials in the "Entry, qualifications and career development" section of this page to complete one pass from problem definition to method and experiment to reproduction statement, then judge whether you are willing to repeat it.
Work pressure, risks and occupational boundaries
Direct answer: the risks this occupation most needs to judge in advance come from three places — an unclear problem definition or unsuitable method making the effort ineffective, experiments and code and data not being reproducible making conclusions untrustworthy, and academic integrity and data and intellectual property and export compliance issues; the drain of long-term uncertainty and failed experiments, unclear interdisciplinary collaboration and requirements, and disputes over authorship and disclosure will amplify these three types of risk.
Tasks and context come from the current O*NET15-1221.00; wages OEWS2025, employment BLS2025—2035; risk control is an editorial suggestion, and the scope of ethics application is explained by jurisdiction using the relevant Chinese documents and U.S. OHRP guidance, with specific IP and export items not verified.
Typical scenario：Starting with a vague objective; not doing a baseline comparison; treating "good results" as a conclusion without a measurable definition.
Roles most affected：
Entry, qualifications and career development
O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The three directions and four stages are an editorial capability path, not a uniform promotion route.
Career outlook and moving to related occupations
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI. OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
United States · 15-1221 national OEWS2025 wages and the BLS2025—2035 projection
BLS Table 1.2 shows 21.8% growth for this occupation in 2025—2035; OOH shows the rounded 22%, not an AI causal study.
Daily work is mainly problem definition and theoretical modeling, system and tool implementation and experimentation, and team and research organization and output: analyze problems to develop solutions involving computer hardware and software; apply theoretical expertise and innovation to create or apply new technology; conduct logical analysis of business and scientific and engineering and technical problems and build mathematical models for computer solution; design computers and the software that runs on them; maintain network hardware and software and direct security measures and monitor availability; participate in interdisciplinary projects such as virtual reality and human-computer interaction or robotics; assign tasks and set goals and performance standards and evaluate work; direct the daily operations of a department and coordinate projects with other departments; participate in staffing decisions and supervise the training of subordinates; and approve and monitor budgets. The page's work context shows 100% use email every day, 94% work indoors in an environmentally controlled setting every day, 69% have face-to-face discussions every day, 62% have sustained contact with others, and 54% have a great deal of decision freedom.
Choose a real task from this page for a small practice project or a conversation with someone in the field. Use the planning guides to set a date, an output and feedback questions before deciding whether to continue.
Conditions for pausing before heavy investment: you have not yet completed one full loop of "problem definition and literature review — mathematical modeling and algorithm or system design — implementation and experimentation — result verification and reproduction — writing and peer review," and you have not verified the mandatory requirements of your institution and funders for research ethics, data and code management, intellectual property and technology export; this occupation's preparation level is Job Zone Five (extensive preparation), and the page's education distribution is expressed in three tiers: respondents said "some positions require a bachelor's degree," "some positions require a doctorate" and "some positions require a master's degree" (the page uses these three tiers and gives no percentages); the occupation page marks this code as Bright Outlook, but this page does not infer individual hiring probability from that.
One-sentence verdict: Computer and Information Research Scientists are not doing "business coding"; they conduct basic computer and information science research as theorists, designers or inventors, and develop solutions to problems in the computer hardware and software field; O*NET defines the occupation as "conduct research into fundamental computer and information science as theorists, designers, or inventors; develop solutions to problems in the field of computer hardware and software," and its tasks rank "analyze problems to develop solutions involving computer hardware and software" and "apply theoretical expertise and innovation to create or apply new technology (such as adapting principles for applying computers to new uses)" as the top two; this page follows the exact occupational scope of O*NET-SOC 15-1221.00.
Signals for exploring further: you are willing to live with open problems and uncertainty over the long term, and you treat "conclusions must be reproducible by peers" as a basic requirement; O*NET work context shows 100% use email every day, 94% work indoors in an environmentally controlled setting every day, 69% have face-to-face discussions every day (31% more than once a week), 64% have telephone conversations every day, 62% have sustained contact with others, 54% sit continuously or almost continuously (45% more than half the time), 54% have a great deal of decision freedom (37% limited), 51% consider work with a work group or team extremely important (43% very important), 48% have a great deal of task autonomy (24% some autonomy), 42% work more than 40 hours per week (58% at 40 hours), 34% consider the consequences of error very serious (24% serious), 12% consider their decisions very important to colleagues or company results (46% important), 58% of the work is moderately automated (25% not at all automated), and 100% never come into contact with disease or infection or whole body vibration or climbing.
The tasks of this occupation can be editorially grouped into three sides: problem and theoretical modeling, system and tool design, and team and resources. The Core items in order include analyzing hardware and software problems, theoretical innovation, assigning tasks, negotiating and coordinating, designing computer software, logical analysis and mathematical models, feasibility, interdisciplinary projects, requirement negotiation, organizational goals and policies and performance; the Supplemental items include network and security availability, department coordination, staff training and budget. The current list does not include "research findings and the structural and functional trends of jobs and occupations" or "developing methods for selection, promotion, evaluation and training," so these are not included as tasks of this occupation. O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks.
Conduct basic computer and information science research as a theorist, designer or inventor, and develop solutions to hardware and software problems
Editorial output check: the problem, method, code experiments and limitations are traceable; the specific disclosure, reproduction and review methods are verified by project and within the legally shareable scope, and each round of peer scrutiny is not treated as a uniform delivery condition.
The main counterparts among the eight neighboring occupations are respectively: planning, coordination and management of information processing, systems analysis and programming activities; management of architectural engineering and its research and development activities; management of life, physical, mathematical and statistical and related research and development; solving scientific, engineering and business data processing problems, systems integration and improvement of existing systems; enterprise database and warehouse architecture, models, standards and integration; data analysis applications, mining, modeling, machine learning and interpretation of results; creating, modifying and testing program code and scripts to specifications; solving complex application, system administration or network problems and systems integration. Programmers and software developers are classified differently, and the fixed neighboring occupation on this page is computer programmers. Tasks may overlap, and this does not prove levels of responsibility, degrees of innovation or automatic mutual recognition.
| Dimension | Primary track | Comparison track |
|---|---|---|
| Core purpose | The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee. |
Core items 1, 6 and 7; editorial grouping, the task pool is not required of every position.
Core items 2 and 8; editorial grouping, the task pool is not required of every position.
Core items 5 and 9; editorial grouping, the task pool is not required of every position.
Supplemental item 1; editorial grouping, the task pool is not required of every position.
Core items 3, 4, 10 and 11 and Supplemental item 2; editorial grouping, the task pool is not required of every position.
Supplemental items 3 and 4; editorial grouping, the task pool is not required of every position.
Editorial six-item verification: problem and method, experiment version, limitations, data and applicable ethical conditions, authorship and disclosure of results and specific IP and export rules; verify the actual jurisdiction, applicable matters and institutional conditions first, and do not treat everything as uniform prior approval.
Check one: are the research question and hypothesis clearly defined and testable
Check two: does the method or system design have a theoretical or empirical basis
Check three: are the experiments and code and data reproducible with version records
| Architectural and engineering managers | Mainly engineering project management and acceptance records. | Architectural and engineering managers coordinate the schedule, budget, people and compliance of engineering projects; this occupation mainly produces research problems and methods in the computer and information field. | People willing to coordinate engineering projects and teams suit that direction; people willing to do theory and method research suit this occupation. |
|---|
| Natural sciences managers | Mainly natural science project management and experimental platform records. | Natural sciences managers coordinate natural science teams and projects and their compliance; this occupation's research object is the principles and systems of computing and information, not natural science phenomena themselves. | People willing to coordinate natural science teams and projects suit that direction; people willing to study computing and information principles suit this occupation. |
|---|
| Computer systems analysts | Mainly requirements and solutions and implementation coordination records. | Computer systems analysts mainly translate business requirements into system solutions and coordinate implementation; this occupation mainly proposes and verifies new theories or methods and technical solutions, and does not carry responsibility for business system delivery. | People willing to do requirements analysis and system solution coordination suit that direction; people willing to do principle and method innovation suit this occupation. |
|---|
| Database architects | Mainly data model and database design and governance records. | Database architects mainly design and govern databases and data models and storage solutions; this occupation addresses broader computer and information science research problems and may involve hardware and algorithms and systems. | People willing to do data model and storage and governance design suit that direction; people willing to do principle and method research suit this occupation. |
|---|
| Data scientists | Mainly data modeling and analysis methods and business delivery records. | Data scientists mainly deliver data-driven modeling and analysis, aimed at business or product conclusions; this occupation focuses more on the novelty and reproducibility of the method itself and may extend to hardware and systems and theory. | People willing to deliver business conclusions through data modeling suit that direction; people willing to study new methods and new principles suit this occupation. |
|---|
The six directions share some capabilities with this occupation: computing and data foundations, problem decomposition and modeling, engineering implementation and verification, and written and oral expression. The real difference lies in the objective and responsibility: the management direction aims at coordinating people and budget and projects; the systems analysis direction aims at transforming requirements into solutions and delivering them; the database architecture direction aims at data models and storage governance; the data science direction aims at data-driven modeling and business delivery; while this occupation centers on basic research and the study and verification of new principles and new technology, with papers and methods and reproducible results as its main output. When choosing, it is advisable to first validate with one deliverable exercise in "problem definition — method and experiment plan — results and limitations statement," rather than comparing titles first.
Moving from this occupation to an adjacent direction usually requires filling in new competencies and qualifications: the department management direction requires people and budget and compliance management experience; the engineering project management direction requires engineering schedule and acceptance management experience; the natural sciences management direction requires project coordination experience in the natural sciences; the systems analysis direction requires requirements analysis and business modeling and implementation coordination experience; the database architecture direction requires engineering experience in data models and storage and governance; the data science direction requires data modeling and business delivery experience. Computing and modeling and verification abilities can transfer, but the objectives and responsibility boundaries of each direction must be rebuilt.
Editorial ratings use the existing occupational responsibilities and AI task analysis on this page. They are not adoption rates, measured effects, job-disappearance or personal-unemployment probabilities, or official scores from Stanford or another external institution. No measured time weights or future robot deployment are assumed.
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI. This is not attributed to AI on that basis.
| Work area | Task | Current change | Human control point |
|---|---|---|---|
| Problem definition and theoretical modeling | Analyze problems to develop solutions involving computer hardware and software and build mathematical models | Literature and formula and derivation drafts can be assisted; the problem definition and modeling judgment remain with people | Confirm that the research question and metrics and data scope and ethics and export bottom line have been written down. |
| System and tool implementation and experimentation | Implement prototypes and conduct experiments and verify and reproduce results | Code and configuration drafts can be assisted; the judgment of experimental validity and reproduction remains with people | Confirm that the experimental design and metrics and baseline and reproduction requirements are specified. |
| System and tool implementation and experimentation | Maintain network hardware and software and direct security measures and monitor availability | Monitoring and log organization can be assisted; security measures and availability judgments remain with people | Confirm that the responsibility boundaries and authorization for network and security and operations are written down. |
| Team and research organization and output | Assign tasks and set goals and performance standards and participate in staffing decisions and budget management | Scheduling and budget table organization can be assisted; personnel and resource decisions remain with people | Confirm that the authorization for team division of labor and performance standards and budget and staffing decisions is specified. |
| Team and research organization and output | Participate in interdisciplinary projects and negotiate computing and system requirements with users and vendors | Communication materials can be assisted; requirement confirmation and collaboration commitments remain with people | Confirm that the requirements and interfaces and agreements for interdisciplinary and external collaboration are written down clearly. |
| Team and research organization and output | Write and publish and report the method and results and limitations truthfully | Drafts and charts and reply drafts can be assisted; responsibility for conclusions and authorship remains with people | Verify whether the target paper, patent or technical disclosure has applicable approval, ethics, confidentiality or publication requirements, and complete the corresponding process only where the actual institution requires it; do not presume approval for all projects. |
| System and tool implementation and experimentation | Manage data and code as the institution and funders require and keep reproducible records | Checklist and log organization can be assisted; compliance judgments remain with people | Confirm that the storage and sharing and access authority for data and code meet institutional and funder requirements. |
The five evidence cards each explain the source, the research object, the conclusion and the limits of use. They answer "what evidence exists," not "will this occupation be replaced."
The three directions are an editorial summary of the task structure; tool uses are still to be verified and do not mean they are already adopted or have changed the whole industry.
AI primarily changesSearch and summarize literature and generate drafts of solutions and modeling and candidate evaluation metrics.
Human responsibility remainsJudge whether the problem is testable, whether the metrics and baseline hold up, whether the conclusions are supported by evidence, and be accountable for the research question and conclusions.
AI primarily changesGenerate drafts of code scaffolding and experiment configurations and data processing and result organization.
Human responsibility remainsJudge whether the experimental design is valid and whether reproducibility is satisfied, and be accountable for the truthful reporting of results and limitations and uncertainty.
AI primarily changesOrganize drafts of papers and reports and charts and review replies and project and budget materials.
Human responsibility remainsConfirm that citations and authorship and disclosure and ethics and export review are completed as required, and be accountable for public release and compliance outcomes.
The five steps below are the editorial process presented on this page, used to explain "who is responsible at which step"; they are not a mandatory process specified by the sources, and do not replace the ethics and data and disclosure requirements of the institution and funders themselves.
Clarify the research question and hypothesis and evaluation metrics and available data scope and your own authority, and the bottom line for ethics and data and export and escalation.
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
Review by yourself together with a supervisor or collaborator or compliance position: whether the problem and metrics hold up, whether the experimental design and reproducibility are satisfied, and whether citation and authorship and disclosure are compliant.
Document the experimental environment and dependencies and versions and randomness and results and deviations; manage data involving humans and partner data according to the minimum necessary and authority requirements.
Research signing, ethics decisions and external disclosure authority are verified against the actual division of labor and rules; the specific responsibilities of tool suppliers and users are judged separately.
Automatically generated literature summaries or code may contain incorrect citations or non-reproducible implementations; adopting them directly without verifying the original text and running validation leads to academic and engineering risk.
Control: Rely only on the original literature and runnable code and raw data; for uncertain points, first go back to the original materials and retain the verification process and conclusion.
First distinguish public literature, restricted materials, personal information and trade secrets, and verify the tools according to actual authority and purpose; do not subject all research data to the same privacy rule. Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority.
Control: Using fictitious or approved materials, verify tool authority and the processing purpose and then de-identify as required; de-identification does not automatically authorize upload, sharing or conducting research.
Editor's version scenario: not recording the tool, model or input data version may hinder reproduction of research-assistance output; this does not mean every experiment is necessarily invalid.
Control: Record the tool model identifier, version, time, input materials and environment; verify experimental validity and reproduction conditions under the applicable research method, and do not prove a conclusion is correct from version records.
Drawing conclusions directly from automatically aggregated results may conceal differences in sample and baseline and method.
Control: Academic conclusions are verified by the corresponding authors and responsible personnel; peer review, publication and institutional review apply according to the actual project and submission system, without a uniform precondition for all research.
When the process is split into "the tool produces a conclusion, a person clicks confirm," a gap easily arises in which no one is accountable for the research conclusion and academic integrity.
Control: Specify the executor, reviewer and approver of each step in the process, and make the records able to answer "who made this judgment and on what basis."
First build solid foundations in mathematics and algorithms, experiment and reproduction methods, and code and data management, then learn to use generative tools to assist with literature and code and documentation; do not treat tool output directly as a research conclusion or experimental evidence, and do not skip verification against the original literature and data.
Use tools for literature search and summaries and solution drafts and code comment organization, but keep the problem definition and hypothesis and evaluation metrics with yourself, and be accountable for the accuracy and verifiability of citations.
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
It cannot be inferred that way. The core tasks of this occupation include problem definition and theoretical modeling, experimental and system implementation and reproduction verification, and academic judgment about conclusions and limitations; these steps rely on professional training, academic responsibility and institutional authorization.
Source: O*NET 15-1221.00 tasks and work activities; ILO exposure indexOriginal source ↗
O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. HotTechnology currently lists 76 U.S. 2025 hiring software keywords, including Python66%, AWS35%, PyTorch29%, Azure25%, TensorFlow25%, SQL20%, Kubernetes20%, Java15%, Docker15%, C++14%; some are below 1% or Not available, and not all 76 give percentages. Hiring mention rates are not software adoption rates, AI usage rates, proficiency or research ability. This round has no evidence of an income premium for software or AI skills.
Source: O*NET 15-1221.00 tasks and work activities; O*NET technology list 15-1221.00Original source ↗
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
Source: O*NET 15-1221.00 tasks and work activities; ILO exposure indexOriginal source ↗
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Source: O*NET national employment trends page (15-1221.00, query entry)Original source ↗
Tool output cannot replace research judgment or existing responsibility; the obligations of researchers, institutions and tool suppliers are verified separately under the actual rules and facts, and suppliers are not said to bear no responsibility at all.
Source: O*NET 15-1221.00 tasks and work activities; institutional ethics and data and export requirementsOriginal source ↗
No reliable public source supports setting a disappearance year for this occupation; this page provides no such prediction. It is more useful to track changes in research paradigms and computing platforms and data rules than to wait for a date.
Source: Editor's note for this pageOriginal source ↗
Suggested priority order: first build solid foundations in mathematics and algorithms, experiment and reproduction methods, code and data management and academic writing, then learn to use generative tools as aids; at the same time limit the use of generative AI to drafts of literature and code and documents and charts, and do not use it to replace research questions, judgments about experimental validity, or responsibility for conclusions and authorship.
Source: O*NET technology list 15-1221.00 (technology keywords); NIST AI RMF 1.0Original source ↗
Editor's trial candidates: use fictitious or approved literature, code and data to assist search and organization, prototyping and writing drafts; check citations, code experiments, omissions and total review time against the originals. This round obtained no research on generative AI reliability, actual adoption rates, efficiency gains or job-loss causation for this occupation, so no claim is made that verified change has occurred.
Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping. Broad background cannot serve as this occupation's monthly pay, percentiles or starting pay.
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
Posting quotations are limited to a specific position, period and pay structure and are not national occupational statistics; this round has no representative recruitment sample.
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
| Educational background | Position direction | Notes and verification suggestions |
|---|---|---|
| A graduate degree in computer science or a related discipline | Research positions in universities and research institutions, and research positions in corporate research institutes and laboratories | For China, a graduate degree or other conditions must be verified against the target institution's current posting; the typical U.S. master's degree is not a threshold for all positions in China. |
| Participation in research projects and output of papers and code and experiment records | Algorithm and systems research direction, data and intelligence and networks and security research direction | Editorial summary: different research directions value the corresponding method abilities and output records more; this page does not map education directly to an income range. |
| Professional certification and compliance and ethics and data management training | Research project management direction, data governance and ethics review support direction | Whether certification, training and research experience are necessary must be verified by target position; the 16 associated certification catalog does not prove a uniform Chinese position requirement. |
| Requirements for data involving humans and privacy and intellectual property and export | Achievement transfer and intellectual property and compliance support direction | Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. |
For China this page provides the broad background of 155491 for professional and technical personnel in enterprises above the designated size and 106080 yuan/person/year for all employed persons; no precise occupational wage is provided.
Editorial summary: mainly research projects and papers and talent development, usually scheduled according to project and assessment cycles.
Editorial summary: mainly applied research and methods and prototype delivery, delivered according to product and project rhythm.
Editorial summary: mainly specialized research and platforms and security support, carried out according to policy and compliance requirements.
Generative tools do not provide the precise wage evidence missing on this page; this round did not verify a causal relationship of tools to changes in statistical classification.
| The 50th percentile | 140,300 USD/year (2025 national median) | Statistical percentiles, not personal monthly income or an offer | The median is 140,300 USD/year, not the mean or an individual wage. |
|---|
| The 90th percentile | 230,630 USD/year (2025 90th percentile) | Statistical percentiles, not personal monthly income or an offer | The 75th percentile is 188,700 USD/year, not an individual ability level. |
|---|
Pay · 10th percentile：82,200 USD/year (2025 10th percentile)；OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD； corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer； the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Pay · 25th percentile：103,570 USD/year (2025 25th percentile)；OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD； corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer； the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Pay · median：140,300 USD/year (2025 median)；OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD； corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer； the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Pay · 75th percentile：188,700 USD/year (2025 75th percentile)；OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD； corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer； the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Pay · 90th percentile：230,630 USD/year (2025 90th percentile)；OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD； corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer； the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
The official hourly wage cannot derive monthly income, after-tax pay or personal annual pay without individual hours and arrangements.
OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Compare by actual institution, duties, region and pay composition; there is no personal income prediction usable on this page.
| Industry | Median wage for this occupation in the industry | How to understand this value |
|---|---|---|
| Software publishers | 211,270 USD/year (2025 industry median for this occupation) | The industry median annual wage for this occupation, not the wage of all publishing workers. |
| Computer systems design and related services | 162,760 USD/year (2025 industry median for this occupation) | The industry median annual wage for this occupation, not the 40% broad industry employment share. |
| Physical, engineering and life sciences research and development | 141,830 USD/year (2025 industry median for this occupation) | The industry median annual wage for this occupation, not the wage of all researchers. |
| Federal government (excluding the postal service) | 124,870 USD/year (2025 industry median for this occupation) | The industry median annual wage for this occupation; the 31% government employment share is not wages. |
Institution, duties, region and pay composition can serve as editorial comparison dimensions; national data do not represent a single position, and this round has no separate effect size for these factors.
42% work more than 40 hours a week, 66% face strict deadlines at least once a week but not every day, and 34% consider the consequences of error very serious; these are respondent distributions and cannot support a quantified causal claim that project review significantly affects the pace.
The current O*NET industry composition is professional, scientific and technical services 40%, government 31%, information 10%, other 19% (including education); this is employment structure, not wages or hiring counts. The OOH detailed industry employer shares and this broad classification are explained separately and are not merged. OOH2025 gives the median annual wage for this occupation in state colleges and universities and professional schools as 85,460 USD; the other four industries are in the table. This round did not obtain regional, form or level breakdowns.
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900； the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900； the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900； the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Which Holland interest type suits computer and information research scientists better?
The interest code on the O*NET page for this occupation is IC (Investigative, Conventional): the interest scores published on the page are Investigative 100, Conventional 63, Realistic 46, Artistic 33, Enterprising 25, Social 17. Investigative corresponds to investigating principles and proposing and verifying hypotheses, and Conventional to working by method protocols and records and versions; the work activities for this occupation are led by using computers, obtaining information, making decisions and solving problems; the interest code describes occupational interest characteristics, not a personal fit conclusion.
Do not use it this way: The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity.
Test my Holland occupational interestsWhich Big Five traits are more relevant to this job?
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
Do not use it this way: Caution: adjusting metrics or selectively presenting experiments to obtain "better-looking" results directly amplifies risk in peer review, reproduction and academic integrity.
View my Big Five personality traitsWhich MBTI type suits computer and information research scientists best?
Interest orientation: the occupation page's interest code is IC (Investigative, Conventional), and the daily tasks are mainly problem definition and theoretical modeling, system and tool implementation and experimentation, and team and research organization and output.
Do not use it this way: This page gives no admission or screening advice based on MBTI type; this round obtained no occupational predictive validity, and specific postings are verified separately.
Understand my MBTI preferencesWhich Enneagram type suits computer and information research scientists better?
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
Do not use it this way: Caution: Enneagram interpretations easily slide into personality diagnosis, and this page does not use it to judge fit.
Explore my Enneagram patternIf I am not good at IQ tests or math, can I be a computer and information research scientist?
This occupation has a clear substantive requirement for computing and mathematics and engineering knowledge: the O*NET knowledge list is led by computers and electronics 89, mathematics 77, engineering and technology 74, English language 72, administration and management 58, design 57; the ability list is led by deductive reasoning and inductive reasoning (both 78), oral comprehension and expression (both 75), fluency of ideas and problem sensitivity and written comprehension (all 72); the skills list is led by critical thinking 75, active listening and reading comprehension (both 69), active learning and oral expression (both 63), mathematics and writing (both 60).
Do not use it this way: Caution: relying on personal memory over the long term without recording the environment and dependencies and data and results will expose risk during reproduction, review and team handover; fixing the process and the documentation is the sustainable approach.
Understand my intelligence and reasoning performanceDo computer and information research scientists need high emotional intelligence?
Communication context: email 100% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; 51% consider teamwork extremely important and 43% very important; external communication 49% extremely important and 33% fairly important, coordination 25% extremely important and 27% very important. Respondent distributions are not a requirement for all positions or a personal ability standard.
Do not use it this way: Caution: omitting reproduction verification or deleting unfavorable results to meet a submission or project-closure deadline creates both academic integrity and engineering rework risk.
Understand my emotional intelligence abilityFits better when: Suitability signal: willingness to coordinate the people and budget and projects and compliance of the information technology department.
Watch for: Main counterpart: planning, coordination and management of information processing, systems analysis and programming activities; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Computer and information systems managersFits better when: Suitability signal: willingness to coordinate the schedule and budget and people and acceptance of engineering projects.
Watch for: Main counterpart: management of architectural engineering and its research and development activities; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Architectural and engineering managersFits better when: Suitability signal: willingness to coordinate natural science teams and projects and experimental platforms and compliance.
Watch for: Main counterpart: management of life, physical, mathematical and statistical and related research and development; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Natural sciences managersFits better when: Suitability signal: willingness to translate business requirements into system solutions and coordinate implementation.
Watch for: Main counterpart: solving scientific, engineering and business data processing problems, systems integration and improvement of existing systems; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Computer systems analystsFits better when: Suitability signal: willingness to do the design and engineering of data models and storage and governance.
Watch for: Main counterpart: enterprise database and warehouse architecture, models, standards and integration; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Database architectsFits better when: Suitability signal: willingness to deliver business or product conclusions through data modeling.
Watch for: Main counterpart: data analysis applications, mining, modeling, machine learning and interpretation of results; tasks may overlap, and the definition does not rule out innovation or computing research, nor imply levels of responsibility, scope or transition difficulty.
Explore related career: Data scientistsNo. Reducing an occupation to "the one most suitable personality" conceals real differences: the way research is conducted in universities and research institutions differs greatly from corporate research institutes and government laboratories. A more useful approach is to validate with one deliverable problem and method and experiment exercise.
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
Yes. Extraversion itself is neither an advantage nor a disadvantage; what matters is whether you can explain the method and results clearly and accept challenge; if you prefer collaboration and exchange, you can focus on interdisciplinary projects and collaboration and talent development tasks.
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
IC is the editorial abbreviation of the top two current occupational interests: Investigative 100, Conventional 63; it describes occupational activity orientation, not the proportion of practitioner types or a personal competence score.
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
This occupation places prominent demands on mathematics and logical reasoning: the O*NET knowledge list is led by mathematics 77 and computers and electronics 89, the ability list is dominated by deductive and inductive reasoning (both 78) and information ordering 69, and the skills list includes mathematics 60; modeling and algorithms and experimental design require reliable quantitative and logical ability.
34% consider the consequences of error very serious, 66% face strict deadlines at least once a week but not every day, and 42% work more than 40 hours a week; these are respondent distributions. Experiment records, reproduction and limitations statements are editorial behavior exercises, and no emotional intelligence score or personality validity is inferred from the context.
This round obtained no reliable validity for Big Five, MBTI, Enneagram or introversion-extraversion predicting performance in this occupation; fictitious literature, code and experiment descriptions can be used for behavioral self-checks, and personality does not decide whether one can enter the field.
Judge by deliverables: can you write a hypothesis and evaluation metrics and baseline for a fictitious problem; can you write an experimental design and reproduction conditions for a fictitious method; can you write a one-page research note including conclusions and limitations and uncertainty. Whichever type of task you are willing to repeat is the direction more worth trying.
Possible consequence：A mismatched problem or method may weaken the conclusion and project progress; the specific impact depends on the hypothesis, evaluation and actual records, and it is not guaranteed that long-term investment will necessarily fail.
How to reduce risk：Write the problem as a testable hypothesis and specify the evaluation metrics and baseline; before starting, verify the feasibility of the method with a small-scale experiment or theoretical analysis; when necessary, review with a supervisor or collaborator before expanding the investment.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
Typical scenario：Only runnable on a personal computer; not recording data processing steps; changing the pipeline without explanation when results are unsatisfactory.
Roles most affected：Positions responsible for experimentation and implementation and reproduction
Possible consequence：Insufficient records of versions or randomness may reduce reproducibility; fixing a seed does not guarantee full reproduction, and the environment and randomness limitations must be stated.
How to reduce risk：Fix the environment and dependencies and random seeds and data version and keep runnable code and configurations; manage data and code as the institution and funders require and provide reproduction notes; report the sources of deviation and uncertainty truthfully.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
Typical scenario：Using others' results as background without citation; listing as an author without confirming contribution; publicly disclosing partner data or technical details that may be controlled without approval.
Roles most affected：Positions responsible for writing and publication and disclosure
Possible consequence：Improper citation, authorship or disclosure may involve research integrity and rights disputes; whether it constitutes academic misconduct, infringement or export violation requires specific rules and facts, and is not asserted wholesale.
How to reduce risk：Complete approval and disclosure as the institution and funders require before submission and public release, and verify citations and authorship; consult the compliance position first for technology or data that may be subject to export control; retain approval and disclosure records.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
Typical scenario：Editorial risk examples: conducting an activity requiring approval without completing the applicable review; disclosing identifiable participant materials without authority; de-identification does not automatically make sharing lawful either.
Roles most affected：Positions responsible for handling data involving humans and privacy
Possible consequence：Failure to fulfil applicable ethics or data protection conditions may create risk; first judge the jurisdiction, activity, exemptions and institutional policy, and do not treat all human-related research as uniformly requiring review and consent.
How to reduce risk：Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
Typical scenario：Substituting oral consensus for written agreements; not documenting requirement changes; not defining partner data and interface authority clearly.
Roles most affected：Positions responsible for team and project organization and outreach
Possible consequence：Misunderstanding of partners and requirements may affect delivery and schedule; the specific consequences depend on project evidence, and no inevitable causation is asserted.
How to reduce risk：Write the objectives and interfaces and deliverables and acceptance criteria clearly in writing before starting; confirm requirement changes per procedure and update the plan; retain meeting and confirmation records.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
Typical scenario：Hiding negative results; modifying data or the pipeline under pressure; enduring alone without seeking review and support.
Roles most affected：Positions responsible for coping with long-term uncertainty and failed experiments
Possible consequence：Repeated failure may bring pressure; misconduct such as data manipulation is not inferred directly from failure or psychological pressure.
How to reduce risk：Treat failed experiments as information and record the reasons and boundaries of failure; use mentor and collaborator and psychological support resources per institutional policy; keep method and data decisions under pressure reviewable.
Evidence：research-review-onet-details · research-review-onet-context · research-review-ethics-cn · research-review-ethics-ai-cn · research-review-ethics-us · editorial
| Entry-level position | What you mainly do early on | Editor's observation candidates | How to prove it without experience | Difference from other directions |
|---|---|---|---|---|
| Graduate and intern researchers (under guidance) | Complete literature review and data processing and experiment execution and record support under mentor guidance. | Whether records are complete, the environment and versions are fixed, and anomalies are reported promptly. | Prove it with de-identified experiment records and code and mentor feedback. | Research, supervision, management and disclosure duties may overlap; verify by actual division of labor and authorization, and stage names do not automatically exclude responsibility or confer ethics approval authority. |
| Researcher (independent sub-problem) | Independently complete the definition, method design, experimentation, reproduction and writing of a sub-problem. | Whether the method and metrics hold up, whether results are reproducible, and whether limitations are stated. | Prove it with papers or technical reports and runnable code and reproduction records. | Research, supervision, management and disclosure duties may overlap; verify by actual division of labor and authorization, and stage names do not automatically exclude responsibility or confer ethics approval authority. |
| Project or direction lead (interdisciplinary and supervision) | Lead interdisciplinary or larger-scale research projects and supervise graduate students and newcomers. | Project objectives and output, quality of method and reproduction, ethics and disclosure and supervision. | Prove it with project and output and supervision and compliance records. | Research, supervision, management and disclosure duties may overlap; verify by actual division of labor and authorization, and stage names do not automatically exclude responsibility or confer ethics approval authority. |
| Research standards and data governance position | Participate in building research standards and data governance and ethics and training and evaluation systems. | Policy documents and implementation, review and training coverage, reproduction and data sharing practice. | Prove it with policy documents and review and training records. | Compared with front-line research positions, it focuses on standards and governance rather than a single project. |
| Laboratory or institution management position | Responsible for coordinating the people and budget and platform and external collaboration and compliance of the laboratory or department. | Platform and resource indicators, people and budget, collaboration and compliance and safety outcomes. | Prove it with department planning and budget and appraisal and compliance records. | Compared with research groups, it adds responsibility for people and platforms and external commitments. |
The preparation level is Job Zone Five (extensive preparation);
Research, code and experiment notes are editorial capability candidates and are not claimed to be a common threshold for most positions.
Background, qualifications and training are verified by specific position and policy, and are not claimed to be uniformly required for most positions.
Background, qualifications and training are verified by specific position and policy, and are not claimed to be uniformly required for most positions.
Fictitious materials: fictitious research background and related work summary (including one methodological limitation and one disagreement over metric definitions), a set of fictitious synthetic data and baseline results, and a fictitious environment and dependency list.
Fictitious condition: a fictitious data and code management and reproduction and ethics and disclosure process.
Fictitious constraint: fictitious computing power and time and available data scope and approval requirements.
Step two: write the experimental design and reproduction conditions (list the environment and dependencies and randomness and data version and control settings, and explain possible sources of deviation and how to control them).
Step three: write a one-page conclusions and limitations statement (including results and uncertainty, applicable scope and failure cases, and data and code and disclosure arrangements), and explain which judgments must be confirmed by an ethics or compliance position.
Evaluation basis: whether the hypothesis is testable, whether the metrics and baseline are measurable, whether the experiment and reproduction conditions are complete, whether the limitations and uncertainty are truthful, whether the data and code and disclosure arrangements are compliant, and whether judgments beyond authority are identified as requiring authorized confirmation.
Evidence forms include de-identified papers or technical reports, runnable code and reproduction records, experiment and version records, review and approval records, and collaboration and authorship agreement records.
For items without evidence, explicitly write "to be added" and give a plan to fill the gap.
Write out the points you disagree with or that need to be added, and explain the basis.
Point out the parts of the record that cannot be traced back (such as missing randomness or data version or approval) and explain what consequences such gaps bring in reproduction and review and audit.
A boundary summary: the judgments you are and are not allowed to make in this project, and the items requiring confirmation by ethics or compliance or your mentor and collaborators.
| Stage | What may be assessed | Observed evidence | Limits of use |
|---|---|---|---|
| Resume and initial screening | Disciplinary background and research direction, output of papers and code and experiment records, and your role and contribution in projects. | Whether the experience matches the position's research direction, and understanding of method and reproduction and ethics and data management. | Initial screening sees only written information; degree and ethics and export requirements must be verified against institution and funder rules. |
| Interview and research presentation | Explaining and answering questions about the hypothesis and method and results and limitations of your own research, and awareness of the limitations of existing methods. |
Organize the fictitious research problem and related work summary into candidate hypotheses and a list of difficulties.
Write the testable hypothesis and evaluation metrics and baseline and note the disagreements in definitions.
Write the control and environment and data and randomness and evaluation procedure.
Write the environment and dependencies and data version and random seed and run instructions.
Organize the fictitious results and baseline and sources of deviation into a draft conclusion.
Write the applicable scope and failure cases and data and code and disclosure arrangements.
| Qualification / status | Mainly suits whom | Practical role in job seeking | Whether it is a universal entry threshold | Application / registration conditions | Level of investment | Important limits |
|---|---|---|---|---|---|---|
| A graduate degree in computer science or a related discipline | Most entrants | BLS typical master's degree, bachelor's for some positions, verify against the posting | O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. |
Typical scope：Complete literature organization and formal derivation and experimental verification support under guidance.
Readiness evidence：Can present a reviewable derivation and verification record.
Credential boundary：Editor's stage candidates: verify mathematics and theoretical foundations and graduate training by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Make a fixed checklist of the derivation and verification points.
Typical scope：Independently complete the modeling and theoretical analysis and verification and writing of a sub-problem.
Readiness evidence：Can present papers or technical reports and verification records.
Credential boundary：Must have formal modeling and proof or analysis ability.
Next moves：Write the modeling scope and assumptions into a stable version.
Typical scope：Lead theoretical direction projects or cross-field collaboration and supervise graduate students.
Readiness evidence：Can present project output and collaboration and supervision records.
Credential boundary：Editor's stage candidates: verify multi-project experience and academic influence records by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Organize direction experience into a reusable method.
Typical scope：Participate in building disciplinary standards and evaluation and training and curriculum systems.
Readiness evidence：Can present policy documents and teaching and inspection records.
Credential boundary：Editor's stage candidates: verify in-institution duties and long-term experience by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Distill policy key points into reusable templates.
Typical scope：Complete code implementation and experiment execution and data organization under guidance.
Readiness evidence：Can present code and experiment and data organization records.
Credential boundary：Editor's stage candidates: verify programming and experimental foundations by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Make a checklist of the environment and dependencies and record points.
Typical scope：Independently complete prototype implementation and experimental design and result reproduction and deviation analysis.
Readiness evidence：Can present runnable code and reproduction notes and result records.
Credential boundary：Must have system implementation and experimental design and reproduction ability.
Next moves：Write the implementation and reproduction process into fixed steps.
Lead research platforms or cross-team systems research projects and supervise newcomers.
Typical scope：Complete scenario investigation and metric organization and verification and record support under guidance.
Readiness evidence：Can present investigation and verification and record materials.
Credential boundary：Editor's stage candidates: verify domain knowledge and basic research methods by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Make a checklist of the scenarios and metrics points.
Typical scope：Independently complete method adaptation and verification and results and limitations statements for application scenarios.
Readiness evidence：Can present verification reports and technical notes.
Credential boundary：Must have method adaptation and evaluation and expression and documentation ability.
Next moves：Write the application verification process into a stable version.
Lead achievement transfer or external collaboration projects and coordinate compliance and intellectual property.
Able to complete literature review and data processing and experiment execution and record support under the guidance of a mentor or senior researcher, follow standards on version and reproduction and data management requirements, and report and correct promptly when a missing item in the method or records or compliance is found.
Able to independently complete the definition, method design, experimentation, reproduction and writing of a sub-problem, and report the conclusions and limitations and uncertainty truthfully.
Able to lead interdisciplinary or larger-scale research projects and supervise graduate students and newcomers, checking the quality of method and reproduction and ethics and disclosure.
Able to participate in building research standards and data governance and ethics and training and evaluation systems, with the corresponding position authorization.
Employment change rate (query entry)
BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Does not establish：It is a statistical scope for the United States nationally or a selected region and does not distinguish institution type and research direction; it also cannot be used to infer the number of positions in a particular institution or research group.
Global (cross-economy) · Cross-economy, occupation level (not a single country's official statistics)
Exposure ranking and measurement method
Generative AI occupational exposure index
The study provides an occupation-level generative AI exposure index, used to compare the degree of task exposure across occupations; it does not provide forecasts of job gains or losses.
Does not establish：It is an exposure measure rather than an adoption rate or automation rate; the cross-economy scope differs from national statistical scopes and cannot be directly converted into job changes for the United States or China.
Mainland China · Mainland China (no one-to-one equivalent occupational sub-class established)
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
Does not establish：NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
These are capability relationships, not automatic promotions or credential waivers.
Transfers：Shared competencies: computing and systems understanding, technical solution judgment, cross-department communication and project drive.
Build next：Editor's gap verification: tasks for planning, coordination and management of information processing, systems analysis and programming activities may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: project planning, technical judgment, cross-disciplinary collaboration and records.
Build next：Editor's gap verification: tasks for managing architectural engineering and its research and development activities may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: research methods and experimental design, data and record management, teamwork.
Build next：Editor's gap verification: tasks for managing life, physical, mathematical and statistical and related research and development may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: systems and data understanding, solution evaluation, communication and documentation ability.
Build next：Editor's gap verification: tasks for solving scientific, engineering and business data processing problems, systems integration and improvement of existing systems may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: data and systems understanding, performance and governance awareness, solution design ability.
Build next：Editor's gap verification: tasks for enterprise database and warehouse architecture, models, standards and integration may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: statistical and machine learning methods, data processing, experiment and evaluation methods.
Build next：Editor's gap verification: tasks for data analysis applications, mining, modeling, machine learning and interpretation of results may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: programming and systems design, code quality and collaboration tools, performance awareness.
Build next：Editor's gap verification: tasks for creating, modifying and testing program code and scripts to specifications may be shared, and it is not presumed that all researchers lack that ability.
View career profileTransfers：Shared competencies: systems design and performance analysis, cross-module collaboration, documentation and review habits.
Build next：Editor's gap verification: tasks for solving complex application, system administration or network problems and systems integration may be shared, and it is not presumed that all researchers lack that ability.
View career profileComputer and information systems managers coordinate the people and budget and projects and compliance of the information technology department; architectural and engineering managers and natural sciences managers coordinate engineering and natural science teams respectively; computer systems analysts mainly translate business requirements into system solutions and coordinate them; database architects mainly design and govern databases and data models and storage solutions; data scientists mainly deliver data-driven modeling and analysis; this occupation centers on basic research and the study and verification of new principles and new technology, with papers and methods and reproducible results as its main output. Shared competencies do not mean interchangeability: the objectives and responsibility boundaries differ.
It cannot be inferred that way. The core tasks of this occupation include problem definition and theoretical modeling, experimental and system implementation and reproduction verification, and academic judgment about conclusions and limitations; these steps rely on professional training and academic responsibility and institutional authorization. In addition, the ILO exposure index registered in this evidence route is only a task exposure measure, not a replacement rate or a forecast of job losses. The more reliable use of tools is drafts of literature and code and data and documents.
From the task structure, the steps affected earlier are the structured and digitizable ones: literature search and summaries, code scaffolding and boilerplate and debugging assistance, data processing and charts and document drafts, and the generation of some experiment configurations. Affected later are the steps requiring judgment and responsibility: problem definition and hypothesis generation, experimental design and validity judgment, judgments about conclusions and limitations and uncertainty, and ethics and authorship and export review. It must be emphasized: this judgment is an editorial compilation, not an exposure measure or job forecast provided by the sources.
Proposing questions worth researching and writing them as testable hypotheses, judging whether the experimental design is valid and whether results are reproducible, stating limitations and uncertainty truthfully when evidence is insufficient, and bearing responsibility for ethics and authorship and disclosure. These abilities rely on professional training and academic responsibility; replacing any of these steps with tool output makes "who is accountable for research conclusions and academic integrity" unanswerable.
This page gives no fixed wage and employment figures for this occupation. This evidence route registers the U.S. wage and employment module as unavailable and requires that "the page must show the gap rather than use a proxy figure" and that "this round does not transcribe static figures"; its claim states explicitly that "U.S. wages should be queried by region and period through the current local wage page for this occupation (15-1221.00); this round does not transcribe static figures, and regional data must not be treated as an individual offer." This page likewise provides no figure for mainland China: the route registers that scope as a module-level gap (no data that can be safely used) and requires that the gap be shown. If you need data, use the local wage page and the national employment trends page for this occupation, queried by region and period.
The page's education distribution is expressed in three tiers: respondents said "some positions require a bachelor's degree," "some positions require a doctorate" and "some positions require a master's degree" (the page uses these three tiers and gives no percentages), and the preparation level is Job Zone Five (extensive preparation); the common path is graduate study in computer science or a related discipline with output of papers and code and experiment records in research projects. Education, output records, certification and institutional and funder approval play different roles and cannot substitute for one another.
The page provides 16 national certifications (such as Certified Computer Scientist, Certified Function Point Practitioner／Specialist, Certified LabWindows/CVI Developer, Certified in Open Source Intelligence, and others); but a certification is not equal to a degree, an employer appointment, signing authority or research qualifications. Research involving humans also requires ethics review, and research involving controlled technology also requires export and security approval; this page does not present any training or certification as a statutory qualification.
Feasible, but it usually requires systematic additional research training: people with a software engineering or data science or mathematics background can transfer their programming and modeling and experimental abilities, but usually still need to complete graduate-level training or equivalent research training, produce verifiable papers or technical reports and reproducible code, and complete ethics and data and export related training and approval as the institution and funders require.
It depends on whether you are willing to live with open problems and uncertainty over the long term and treat reproducibility as a basic requirement; the preparation level for this occupation is Job Zone Five (extensive preparation), and research output cycles are usually long. Per the route requirement, this page does not provide fixed employment and wage figures for this occupation; for employment projections, use the national employment trends page for this occupation queried by region and period. It is advisable to first use the seven-day exercise on the page to verify whether you are willing to repeat this process.
This evidence route registers mainland China wages as a module-level gap (data status: no data that can be safely used); the scope note states explicitly that "this round no mainland China public source established a national occupational wage distribution corresponding one-to-one with this occupation," and states that "the page must show the gap rather than use a proxy figure"; the handling note is "retain the gap and show the source and occupational mapping boundary." Therefore this page publishes no figure in the mainland China salary module.
This evidence route registers the U.S. wage and employment module as unavailable (data status: no data that can be safely used); its claim states explicitly that "U.S. wages should be queried by region and period through the current local wage page for this occupation (15-1221.00); this round does not transcribe static figures, and regional data must not be treated as an individual offer," and another claim requires that "the page must show the gap rather than use a proxy figure." Therefore this page provides only the query entry and scope notes, and does not fix figures.
The exercise and the seven-day schedule are fictitious materials designed editorially by FermatMind Test to help readers judge whether they are willing to repeat this workflow; they are not research findings or ethics or legal or academic evaluation advice; they must not be used for real research or as a basis for ethics and compliance decisions.
This round verified the nine sections section by section and against citations: mixed-in tasks, context figures and frequency boundaries were corrected, and OEWS2025 wages, the BLS2025—2035 projection, the NBS2025 broad background and the applicable ethics conditions were added. For local author self-review only.
Nine-section verification for this round: 2026-10-09.
O*NET2026 current occupational tasks, context, knowledge, abilities and interests, 2025 hiring keywords, the current 16 certifications and industry composition; BLS2025 wages / 2025—2035 projection and OOH education and industry wages; NBS2025 same-version primary text, China's 2023/2026 science and technology ethics documents and U.S. OHRP scope guidance, explained by use.
Limitations：Supports the precise occupational identity, the 15 tasks, work activities and work context; does not include what any specific jurisdiction or institution requires for research ethics and data and export.
Limitations：Supports occupational characteristics, knowledge, abilities, the interest code IC and work styles (the page gives impact scores); the page marks this code as Bright Outlook.
Limitations：Supports the industry structure (professional and scientific and technical services 44%, government 29%, information 11%, other industries 16%); this is a share structure and does not represent the number of positions or wages.
Limitations：The page lists 76 software skills and their mention shares (Python 66%, Amazon Web Services 35%, PyTorch 29%, Microsoft Azure 25%, TensorFlow 25%, SQL 20%, etc.); mention share does not equal proficiency or research ability.
Limitations：Supports the preparation level (Job Zone Five: extensive preparation) and the page's three-tier education distribution statement (some positions require a bachelor's degree / doctorate / master's degree, with no percentages on the page); the local training page failed to be retrieved this round (SOURCE_RESPONSE_TOO_LARGE) and is not used as a source.
Limitations：The page lists 16 national certifications (Certified Computer Scientist, Certified Function Point Practitioner／Specialist, Certified LabWindows/CVI Developer, Certified in Open Source Intelligence, and others); a certification is not equal to a degree, an employer appointment, signing authority or research qualifications.
Limitations：This evidence route registers U.S. wages as unavailable (us-salary-static-unavailable) and requires that "the page must show the gap rather than use a proxy figure" and that "this round does not transcribe static figures"; this page provides only that query entry and scope notes and publishes no fixed figures.
Limitations：This evidence route registers U.S. employment as unavailable and requires that static figures not be transcribed; this page provides only that query entry and scope notes and publishes no employment size or growth rate or average annual openings figures; projections contain assumptions and error.
Limitations：Based on such scopes, this evidence route registers the mainland China wage for this occupation as a module-level gap (data status: no data that can be safely used) and requires that "the page must show the gap rather than use a proxy figure"; this page publishes no figure for mainland China.
Limitations：It is the mainland China labor relations and wage information entry registered in this evidence route; this round did not establish a national occupational wage distribution corresponding one-to-one with this occupation on that basis, and this page publishes no figure from it.
Limitations：Supports the general framework of mainland China occupational classification; this page does not use it to determine this occupation's Chinese equivalent sub-class or wage scope.
Limitations：It is an exposure measure, not an adoption rate, automation rate or forecast of job losses; cross-economy scopes cannot be directly converted into national job changes.
Limitations：Editorial compilations and fictitious materials, not research findings or industry facts; must not be used for real research or as a basis for ethics and compliance decisions.
Limitations：It is a framework for AI risk governance and management processes and contains no occupation-level adoption data; citing it does not constitute a quantified conclusion about this occupation's risk.
Limitations：Per the route requirement this page does not transcribe static wage figures; query results must retain the region and period and statistical scope, and regional data must not be treated as an individual offer.
Limitations：Supports the percentage distributions of work context (environment and protection, communication and collaboration, autonomy and decisions, precision and responsibility, posture and activity); they are respondent distributions, not a guarantee for an individual position.
Limitations：Supports the coverage scope of wage statistics, the items included and excluded, and the standard work-year conversion definitions; not to be used to infer personal income.
Limitations：It is the source scope and query entry for the U.S. national employment projection; per the route requirement this page does not transcribe static projection figures for this occupation, and projections contain assumptions and error.
Limitations：Supports the percentage distributions of work context (environment and protection, communication and collaboration, autonomy and decisions, precision and responsibility, posture and activity); they are respondent distributions, not a guarantee for an individual position.
Limitations：O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity. The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee.
Limitations：O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity. The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee.
Limitations：The current O*NET industry composition is professional, scientific and technical services 40%, government 31%, information 10%, other 19% (including education); this is employment structure, not wages or hiring counts. The OOH detailed industry employer shares and this broad classification are explained separately and are not merged.
Limitations：HotTechnology currently lists 76 U.S. 2025 hiring software keywords, including Python66%, AWS35%, PyTorch29%, Azure25%, TensorFlow25%, SQL20%, Kubernetes20%, Java15%, Docker15%, C++14%; some are below 1% or Not available, and not all 76 give percentages. Hiring mention rates are not software adoption rates, AI usage rates, proficiency or research ability.
Limitations：O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity. The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee.
Limitations：Actually read the 16 certification candidates; they are not a uniform threshold, ability validity, or verification of all application conditions.
Limitations：OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Limitations：BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Limitations：NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published.
Limitations：Reuses the same-version 2025 public summary read this round; exposure is not this occupation's adoption rate, reliability, job reduction or disappearance time; no claim is made that the full detailed tables have been verified.
Limitations：The same-version 1.0 framework read is a voluntary governance reference, not evidence of a statutory process or effect for this occupation.
Limitations：OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered.
Limitations：O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity. The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee.
Limitations：Reuses the same-version wage coverage and method read; the self-employed and unincorporated owner-partners are excluded, salaried executives of incorporated companies may be covered, and the statistical hourly wage does not derive personal income.
Limitations：BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Limitations：O*NET15-1221.00 Updated2026 covers basic computer and information research and hardware and software problem solving, with 15 tasks comprising 11Core and 4Supplemental. The task pool covers modeling, system design, interdisciplinary research, requirement negotiation and some management and operations tasks, and not every position does all tasks. O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The current O*NET list has 33 knowledge items, 52 ability items and 21 work styles; inquisitiveness 100 / innovation 94 / attention to detail 83 / adaptability 77 / achievement 75, knowledge computers 89 / mathematics 77 / engineering 74 / English 72, abilities deductive and inductive 78 / oral comprehension and expression 75 / problem sensitivity and written comprehension 72. Occupational indicators are not personal screening test scores and do not prove personality predictive validity. The current 57 context items are respondent answer distributions: email 100% every day, controlled indoor 94% every day, face-to-face 69% every day and 31% at least once a week but not every day, telephone 64% every day; sitting 54% continuously and 45% more than half the time; 42% work more than 40 hours a week, 58% at 40; disease 99% never, whole body vibration and climbing 100% never, heights 91% never, hazardous conditions 73% never. Automation level is 58% moderate, 17% slight and 25% none, which is not an AI adoption rate or task replacement share; distributions are not a personal guarantee.
Limitations：O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. The 2025 industry median annual wages for this occupation are: software publishing 211270, systems design 162760, physical, engineering and life sciences research and development 141830, federal government excluding postal 124870, state colleges and universities 85460 USD. BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI.
Limitations：Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority.
Limitations：Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. Article 37 takes effect from the date of issuance.
Limitations：Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. Official educational guidance; not treated as the full text of the regulations or as verification of specific consent and exemption clauses.
Limitations：Main counterpart: planning, coordination and management of information processing, systems analysis and programming activities; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: management of architectural engineering and its research and development activities; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: management of life, physical, mathematical and statistical and related research and development; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: solving scientific, engineering and business data processing problems, systems integration and improvement of existing systems; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: enterprise database and warehouse architecture, models, standards and integration; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: data analysis applications, mining, modeling, machine learning and interpretation of results; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: creating, modifying and testing program code and scripts to specifications; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Limitations：Main counterpart: solving complex application, system administration or network problems and systems integration; may overlap, and is not evidence of levels of responsibility, innovation ability or transition distance.
Precise Chinese occupational wage mapping, U.S. regional, form and level breakdowns, AI adoption reliability and causal effects, personality validity, certification conditions and fees, and recruitment representativeness are still missing; specific export items were not verified. Old fact sources and identities outside the scope are unchanged, and conflicts are in the author report.
MBTI (English) · Big Five (English) · MBTI (Chinese) · Big Five (Chinese)
| Advance research based on peer-reviewed literature and open problems, mathematics and computer science theory, experiment and reproduction records, and the ethics and data and intellectual property requirements of the institution and funders |
| Main counterparts and basis | Computer and information systems managers coordinate the people, budget and projects of the information technology department; this occupation centers on basic research and the creation and verification of new principles and new technology | Verify each step against the O*NET occupational definition and its 15 tasks, peer-reviewed literature and theory, experiment and reproduction records, and institutional ethics review and data and code management standards |
|---|
| Core process | Systems analysts can solve complex scientific and engineering applications and improve systems, while this occupation leans toward basic research and new technology; the two may overlap, and the definition does not rule out business delivery by this occupation or innovation by neighboring occupations. | Editorial research process candidates: define the problem and the literature → modeling or design → prototype experimentation → verification records → reports or papers and applicable review → iterate. OOH supports experimentation, publication and conference exchange, and not all research must follow the same order or deliver only after peer review. |
|---|
| Main outputs | Data scientists mainly deliver data-driven modeling and analysis; this occupation focuses more on the novelty and reproducibility of the principles and methods themselves and may extend to hardware and systems | Research questions and literature reviews, modeling and design documents, code and data and version records, experiment and reproduction notes, papers and reports and patent materials, review records, system and security operations records |
|---|
| Responsible party | This page presents task and output boundaries and does not constitute any statement of ethical, legal, intellectual property or research evaluation conclusions | Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. |
|---|
| Admission and licensing | NBS2025 (published 2026-05-15): professional and technical personnel in enterprises above the designated size averaged 155,491 yuan/person/year and all employed persons 106,080 yuan/person/year, both cross-occupation pre-tax annual average wage background. Enterprise research positions should be referenced only by actual professional and technical duties and survey unit coverage; the name cannot be used to map all universities or research institutions, and these are not this occupation's median, monthly pay or starting pay. Insufficient evidence: this round did not establish a precise Chinese occupational and national occupational wage mapping, which does not mean China lacks this sub-class or that it was never published. | O*NET JobZone5 is the extensive preparation level; the education responses for this occupation are that some positions require a bachelor's, doctorate or master's degree, without proportions. The typical BLS entry is a master's degree, no related work experience, no on-the-job training; OOH usually requires a master's or higher degree in computing or a related field, some employers prefer a doctorate, and some federal government positions may accept a bachelor's degree. The extensive level's many years of experience cannot be generalized into a uniform five-year or doctorate threshold for all positions. Study and research records are editorial capability candidates and do not guarantee a position. |
|---|
| Conclusion boundaries | Cross-position comparison cannot rest only on similar titles; whether the role aims at basic research and new methods as its output, whether it carries product delivery or department management responsibility, and differences in the type of institution mean that similar tasks do not imply interchangeability | OEWS2025 U.S. national 15-1221 employed annual wages at the 10/25/50/75/90 percentiles are 82,200/103,570/140,300/188,700/230,630 USD; corresponding hourly wages 39.52/49.79/67.45/90.72/110.88 USD. The median is not a mean, starting pay or an individual offer; the self-employed and unincorporated owner-partners are excluded, while salaried executives of incorporated companies may be covered. BLS Table 1.2 for 15-1221: 2025—2035 employment 38,600→47,000 people, net increase 8,400, growth 21.8% (OOH rounds to 22%), average annual openings 2,900; the self-employed share is not provided (—), not 0. Openings include growth and replacements and are not the net increase, current hiring or a personal hiring guarantee, and are not attributed to AI. |
|---|
Check four: are the results and limitations and uncertainty stated truthfully
Check five: are the data and privacy and ethics review involving humans completed as required
Check six: do papers and patents and technical disclosures comply with institutional, funder and export control requirements
Knowledge is dominated by computers and electronics 89, mathematics 77, engineering and technology 74 and English language 72;
Work activities are led by using computers 99, obtaining information 91, and making decisions and solving problems 91.
Computing and experiment platforms and job scheduling and monitoring tools;
HotTechnology currently lists 76 U.S. 2025 hiring software keywords, including Python66%, AWS35%, PyTorch29%, Azure25%, TensorFlow25%, SQL20%, Kubernetes20%, Java15%, Docker15%, C++14%; some are below 1% or Not available, and not all 76 give percentages. Hiring mention rates are not software adoption rates, AI usage rates, proficiency or research ability.
Multiple reasonable approaches are allowed; if you consider some information insufficient to judge on, state explicitly the parts that are insufficient and the additional information needed.
The problem and data and process and authority given in the prompt are fictitious values, used only to practice hypothesis and design and management approach.
A collaboration note: your approach to communication and documentation with mentors and collaborators and the project team and the compliance position.
| Whether the person can explain the method clearly, state limitations truthfully, and distinguish correlation from causation. |
| A presentation is not equal to long-term ability; whether there is a trial research or collaboration assessment is verified by specific position, and is not claimed to be a universal process. |
| Method and code assessment | Write the hypothesis and evaluation metrics and experimental design for a given problem, and explain the reproduction conditions and possible deviation. | Problem definition, metrics and baseline, reproduction design, limitations and ethical considerations. | Assessment materials differ from the complexity of real research; results are for reference and not conclusive. |
|---|
| Probation or collaboration period | Take on a specific sub-problem within the mentor's or research group's scope and undergo method and reproduction and standards checks. | Whether the person records and submits per standards, reports risks promptly, and how quickly they correct after problems are pointed out. | Performance during the collaboration period is not the ceiling of long-term ability; direction and team and resources have a large influence. |
|---|
Answer three questions: whether you are willing to repeat this; which step engaged you most; and which capability you will build next week (problem and modeling, experimentation and reproduction, writing and expression, or data and compliance).
| Verify conditions institution by institution and with each body or certification; this round did not verify all qualification training and examination requirements |
| Insufficient evidence: specific fees and time must be verified by program |
| The page's education distribution is expressed in three tiers: respondents said "some positions require a bachelor's degree," "some positions require a doctorate" and "some positions require a master's degree" (the page uses these three tiers and gives no percentages); this page does not map education directly to income. |
| Participation in research projects and output records | People who carry out research and development work | Research records can demonstrate method and reproduction ability but do not automatically determine project role | No (it is an ability condition) | Verify conditions institution by institution and with each body or certification; this round did not verify all qualification training and examination requirements | Insufficient evidence: specific fees and time must be verified by program | Output records are the foundation for taking on research work, but they are not equal to position authorization and ethics approval authority. |
|---|
| Professional certifications (16 in total, including Certified Computer Scientist) | People who want to demonstrate professional ability | Certification catalog candidates; employer recognition and ability validity were not verified | No (a certification is not equal to a degree or research qualifications) | Verify conditions institution by institution and with each body or certification; this round did not verify all qualification training and examination requirements | Insufficient evidence: specific fees and time must be verified by program | The page provides 16 national certifications; a certification is not equal to a degree, an employer appointment, signing authority or research qualifications. |
|---|
| Institutional and funder ethics and data and export approval | Research involving humans or research involving controlled technology | Determines whether research may proceed and whether results may be published | Verified by specific activity, jurisdiction and applicable regime; not all research requires uniform prior approval | Submit for review and approval as the institution and funders require | Insufficient evidence: specific fees and time must be verified by program | Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. |
|---|
| Research standards and data and code management requirements | All personnel who carry out research | Determines the boundaries of data and code management and sharing | Verified by specific activity, jurisdiction and applicable regime; not all research requires uniform prior approval | Follow standards on version and reproduction and sharing requirements | Insufficient evidence: specific fees and time must be verified by program | Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. |
|---|
| Authorship and disclosure and intellectual property agreements | Personnel who participate in collaboration and publication | Determines the authorship and disclosure method of the output | Verified by specific activity, jurisdiction and applicable regime; not all research requires uniform prior approval | Confirm authorship and disclosure per agreements and policy | Insufficient evidence: specific fees and time must be verified by program | Articles 2, 4, 9 and 15 of China's 2023 Measures for Ethical Review of Science and Technology (Trial) limit the review and risk responsibility for activities involving humans, animals or other higher ethical risk and for the applicable units; Articles 2, 9 and 12 of the 2026 Measures for Ethical Review and Services of Artificial Intelligence Science and Technology (Trial) set out the scope of application and committee or commissioned-service review for AI research and development with potential ethical risk within China. The U.S. OHRP Lesson2 distinguishes human-subjects research subject to the Common Rule, non-research, research not involving human subjects and exemptions, and some exemptions still require limited IRB review; institutional policies may impose additional conditions. It cannot be said that all projects involving humans or all computer projects uniformly require approval or consent. The specific conditions for data handling, informed consent, IP and export must be verified separately by matter, and de-identification does not automatically grant processing or sharing authority. |
|---|
This page does not guarantee that any training or certification is recognized by a particular institution, and does not replace specific judgments on ethics and data and intellectual property and export or legal advice.
Readiness evidence：Can present platform and project and supervision and compliance records.
Credential boundary：Editor's stage candidates: verify platform building and multi-team collaboration experience by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Organize platform and project experience into a reusable plan.
Typical scope：Participate in building engineering standards and data and code governance and reproduction systems.
Readiness evidence：Can present standards documents and inspection and training records.
Credential boundary：Editor's stage candidates: verify in-institution duties and long-term experience by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Organize system-building experience into handover-ready policy text.
Readiness evidence：Can present project and compliance and intellectual property and delivery records.
Credential boundary：Editor's stage candidates: verify collaboration and compliance and intellectual property experience by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Organize transfer experience into a reusable plan.
Typical scope：Participate in building achievement transfer and intellectual property and compliance and disclosure systems.
Readiness evidence：Can present policy documents and inspection and training records.
Credential boundary：Editor's stage candidates: verify in-institution duties and long-term experience by actual position. This round has no representative recruitment sample supporting a uniform threshold for all positions.
Next moves：Organize policy key points into handover-ready text.
Limitations：Editorial ratings use the existing occupational responsibilities and AI task analysis on this page. They are not adoption rates, measured effects, job-disappearance or personal-unemployment probabilities, or official scores from Stanford or another external institution. No measured time weights or future robot deployment are assumed.
