From Human Learning & Judgement to AI Systems

Topic / Focus

So far, we have asked how humans learn, how judgment is shaped by values and social context, and who has the power to define what counts as normal or good. Today we add another question: What is the system actually doing?

Understanding how an AI system works does not settle ethical or political questions. But it can clarify what is being represented, measured, learned, and optimized – and make the stakes of those questions more concrete.

Critical AI literacy means moving between technical mechanisms, social context, values, power, and consequences – while also recognizing what we still do not know.

Slides & Activities

• Slides
• Activity

Guiding Questions

• What is the difference between observing that a system “learns what you like” and explaining how it works?
• Which parts of an AI system can we already analyze with the tools from Topic 1, and which require technical knowledge we do not yet have?
• How does mechanistic detail change a social critique from vague to specific?

Analysis Framework

Dimension	Question
Representation	What information does the system use, and how is it represented?
Procedure	What steps or algorithms transform inputs into outputs?
Learning	What changes through training or experience?
Categories	What counts as similar, different, normal, or anomalous?
Values	What outcome gets rewarded or optimized?
Norms	What counts as correct or acceptable?
Power	Who chose these things, and who is affected?
Impacts	What benefits, harms, risks, or downstream effects follow?