This text explains to students the phenomenon of algorithmic bias and how faulty training data can lead AI systems to produce discriminatory outcomes. It is part of the “KI in meinem Alltag” school material package and links to related interviews and infographics on the same topic.
Why it matters for ENACT-AI
It encourages learners to question AI-generated or AI-based decisions, recognise discriminatory outcomes and reflect on the responsibility of developers and users.
Learning outcomes
Participants will be able to:
explain what algorithmic bias is;
describe how training data can influence AI outputs;
identify examples of discrimination caused or amplified by AI systems;
recognise the relationship between social inequalities and algorithmic systems;
critically evaluate AI-based decisions and outcomes;
explain why diverse and representative data are important;
reflect on ways to identify and prevent discrimination in AI systems.
Suggested activity
“Can AI Be Fair?” – Bias Investigation.
Present learners with several hypothetical AI decisions, such as facial recognition, recruitment or recommendation systems. In groups, they identify who might benefit or be disadvantaged by each system and discuss what types of data or assumptions could create bias. They then redesign the system by proposing measures to make it fairer and more inclusive. Finish with a discussion about whether completely unbiased AI is possible.