Loading page content
Original practice bank · All explanations and revision guides free
Try 20 original AI literacy scenarios on models, tools, verification, privacy and judgement. Free worked answers and four full guides. Untimed; no signup.
20 questions · about 10 minutes · no signup
Read the declared source, tool capability and permission in each scenario, then choose one answer. All examples are invented, not recordings from a named model. There is no timer or speed bonus; around ten minutes is an editorial estimate. All 20 valid choices are required before counts appear, with five items per section. This is a finite knowledge exercise, not a live prompting benchmark, an employment-readiness score or an official AI certification. Interpret any calculator or reference use as part of your practice conditions.
Preparing the test…
All questions and full results are free, with no signup. Saving is optional; only saved attempts can be resumed after a refresh or language change. Tab drafts expire after 24 hours and clear on completion or retake.
What evidence supports an AI-generated claim, and what does a tool actually have permission to do? Choose answers to 20 original scenarios across models and data, prompts and tools, checking outputs, and privacy and judgement. Each item has a keyed answer and worked explanation reviewed against selected primary sources and its explicit conditions. Small arithmetic and JSON examples were checked separately. All four full guides are free. The outputs and situations are invented; this website runs no AI model or connected action and asks for no private text.
A matching key counts as one correct item, and another valid choice counts as zero. Four sections have five items each, giving a total out of 20. All valid choices are needed before counts appear. There is no percentile, speed weighting, personality type, moral rating or employer threshold.
Every section has a public revision guide. Completing the practice shows every guide and item explanation without assigning a type by score. Public guides contain no personal responses or counts; studying before practice is welcome. This is not a secure exam.
Review training and inference, generated content, narrow evaluation, confabulation and evidence for current facts.
Review specific prompts, targeted revision, deterministic calculations, external instructions and evidence of tool actions.
Work through arithmetic, citations, JSON types, held-out evaluation and sample percentages using fresh invented examples.
Review confidential input, responsible decisions, group-level evidence, disclosure rules and authorised tool use.
Twenty original right-answer scenarios cover models and data, prompts and tools, checking outputs, and privacy and judgement. Each section has five items. No coding, model subscription or personal document is needed.
No. Situations and outputs are invented for this knowledge exercise. Small arithmetic and JSON checks were run separately, but no AI model, connected tool action or named-model benchmark is executed here.
No. It counts matching choices on this finite original bank. It is not a live prompting test, employment screening instrument, population ranking or official UNESCO, NIST, OECD or OWASP qualification.
Yes. Every complete valid attempt receives all choices, keys, worked explanations and four full guides at any count. Public guides are readable before starting. No signup, email, payment or certificate purchase is required.
The questions declare their own conditions. Capabilities, retention, updates and permissions can differ by system and setting. Use current primary documentation for an actual tool; a quiz answer does not establish all future behaviour.
There is no timer or speed bonus. Around ten minutes is an editorial estimate. You can choose your practice conditions, including references, and interpret the count accordingly.
Answers stay in the browser. Optional tab drafts expire after 24 hours and clear on completion, exit-and-clear or retake; refresh loses progress without one. Sharing copies only a public practice link, without answers, counts or private identifiers.
Zero, partial and perfect counts receive every prompt, selected answer, key and explanation, followed by all four complete guides. Public guides are readable before starting and contain revision material rather than a personal score. No email, payment or certificate purchase unlocks a withheld answer.
AI systems, connected capabilities and organisational rules differ. The items specify the input, permission or evidence needed for each answer. Examples are invented and no model, email, publication or account action runs here. An apparent productivity or expertise claim in a generated draft is not treated as an observed result.
FermatMind wrote the questions, distractors, explanations and guides. The linked sources support selected concepts; their frameworks, example questions and certification schemes are not reproduced or claimed as this bank’s validation. No vendor endorsement, norm sample or universal skill level is claimed. Repeating a fixed bank after reading its keys can improve recall.
These sources provide background, not validation of this questionnaire or an individual diagnosis.
Website content edition 1.0.0 ·
Source question / scoring / report versions: fermatmind_ai_literacy_original_20_v1 / practice_equal_correct_counts_v1 / fm_ai_literacy_four_full_revision_guides_v1