AIMentor — AI Expertise Learning Platform

13 questions · pass mark 80% · 20 minute limit. The timer auto-submits at zero.

Time remaining: 20:00

Question 1easy
What is AI governance?
A way to make models fasterPolicies, processes, and controls ensuring AI is built and operated legally, ethically, and accountablyA type of fine-tuningA vector database feature
Question 2easy
How does the EU AI Act classify AI systems?
By model sizeBy risk to people's rights and safety, in four tiers: unacceptable, high, limited, minimalBy programming languageBy token cost
Question 3medium
Does the EU AI Act apply to a non-EU company?
No, only to EU-based providersYes — it applies to systems placed on the EU market regardless of where the provider is basedOnly to open-source modelsOnly to models over 100B parameters
Question 4medium
An AI tool that ranks job applicants for hiring falls into which EU AI Act tier?
Minimal-riskHigh-risk (strict obligations: risk management, data governance, documentation, human oversight, conformity assessment)Unacceptable / bannedNo tier applies
Question 5medium
What obligation applies to a limited-risk system like a customer chatbot?
Full conformity assessment and registrationTransparency — users must be told they're interacting with AI (and AI-generated content labelled)It is bannedNo obligation at all
Question 6hard
What is the consultant's correct role on a plausibly high-risk classification?
Render a final legal verdict themselvesSpot that it's plausibly high-risk and escalate to legal/compliance — don't self-certifyIgnore it if the client hasn't askedAssume it's minimal-risk to avoid delay
Question 7hard
Where does bias enter an AI system? (Select all that apply)
Training-data bias reflecting historical/societal inequitiesRepresentation bias — some groups under-represented in the dataYour pipeline — prompts, examples, retrieval corpus, or labels encoding skewThe choice of programming language
Select all that apply.
Question 8hard
How do you MEASURE bias in an AI system?
Trust a single aggregate accuracy scoreDisaggregated evaluation — run the eval set sliced by group and compare scores across groupsAsk the model if it is biasedIncrease the temperature
Question 9medium
Which are transparency/documentation duties for AI systems? (Select all that apply)
Disclose to users when they're interacting with AI and label AI-generated contentProvide model/system cards describing intended use, limitations, data, and evaluationFormal technical documentation for high-risk systemsHide the fact that AI is involved to seem more human
Select all that apply.
Question 10hard
Why is 'the AI decided' not a valid accountability position?
AI decisions are always correctAI is not a liability shield — governance requires a named human owner with real oversight and an audit trailBecause models are open sourceBecause it costs too much
Question 11hard
What is the key insight linking the earlier technical tracks to governance?
Governance requires entirely separate toolingThe technical controls (human-in-the-loop, evals, tracing, tenant isolation) ARE the governance controls, documented and made accountableGovernance only applies to trainingGovernance replaces engineering
Question 12medium
Which EU AI Act tier (one word) BANS practices like social scoring outright?
Question 13hard
What kind of evaluation (one word, describing the slicing by group) reveals bias an aggregate score hides?

0 / 13 answered

Governance & Responsible AI — Final Exam