01 / 20–30 minutes
Assess the evidence.
Separate a headline claim from what the evidence actually establishes. Identify the questions that could change your decision.
Open workspace ↗Building AI policy capacity
Big questions.
Better policy.
Build the judgement to act. Practical tools, cases and free courses for people shaping how AI is governed.
Open resources. Practical questions. Work worth doing.
01 / Use a tool
The policy workbench ↗From a difficult question
to a working brief.
02 / 30–45 minutes
Compare the options.
Develop a defensible recommendation by comparing mechanisms, distributional effects, delivery constraints and uncertainty.
Open workspace ↗03 / 30–45 minutes
Plan meaningful oversight.
Turn ‘human oversight’ into named responsibilities, practical controls, monitoring and a route to pause or stop.
Open workspace ↗Structured prompts, complete worked examples and primary sources. Save drafts, compare your reasoning and export something you can develop further.
02 / Examine a case
Explore the case library ↗Good judgement
takes practice.
Documented case / 20 min
Robodebt: oversight beyond the model
Compare transparency, review rights and independent scrutiny after a welfare-debt inquiry.
Australia · 2023 inquiry recommendations
Examine the case ↗Documented case / 20 min
Singapore’s retinal AI: accuracy is not a health outcome
Read a strong diagnostic result carefully before claiming benefits for a screening service.
Singapore · with external validation cohorts · 2017 diagnostic-validation study
Examine the case ↗Documented case / 15 min
Brazil’s Victor: faster classification, faster justice?
Separate a dramatic task-time claim from the outcomes of a judicial service.
Brazil · 19 August 2026 court account
Examine the case ↗Documented decisions from the UK and beyond, dated primary sources and explicit limits. Examine the evidence, try the decision and compare your reasoning with an AIPEX analysis.
03 / Learn the craft · Free & self-paced
Build a foundation.
Put it to work.
An enduring purpose
Understanding AI.
Building public judgement.
AIPEX is building capacity to understand, develop and deliver effective AI policy. We begin with useful work: careful teaching, questions that can be examined and resources people can put into practice.