Risks & Ethics
AI Bias
Also known as: algorithmic bias
In one line
Systematic errors in AI output caused by imbalanced training data or design.
What does AI Bias mean?
Because LLMs learn from internet text, they inherit historical bias — gender, race, geography, language. Mitigation involves data curation, evaluation, and human review.
A real-world example
A hiring model that ranks male-coded CVs higher because past hires were mostly male.