L·ALiving AI AtlasA Coding As Creating project

TOPIC 01 OF 05

Bias, fairness, and representation

Ask who carries the error
PLAIN-LANGUAGE IDEA

AI systems learn patterns from selected data and are shaped by labels, objectives, product choices, deployment conditions, and feedback. Harm can appear as stereotyping, exclusion, poorer accuracy for some groups, or an apparently neutral process that distributes opportunity unevenly.

SEE IT IN A SITUATION
An image tool repeatedly shows men for ‘technical leader.’ A hiring summary tool may also omit evidence expressed in unfamiliar language. The response is not only to rewrite a prompt, but to test patterns and reconsider whether the tool belongs in the decision.
DO NOT MISS THIS

Fairness is not a single score. Different groups and harms can require different measures, and improving an average can conceal worse outcomes for a smaller group.

TRY THIS NOW

Ask: Who is represented? Who is missing? Who is misread? Who experiences the consequence? Who can challenge it?

CONNECTED LANDMARKS
Next idea →
◉
Important topics change.

Environmental impact, copyright, privacy, security, and EU obligations require dated sources and context. The Atlas teaches durable decision habits and clearly separates them from legal or professional advice.