CKO-027Hallucinations, Bias & FairnessStrong evidence

Can AI amplify bias?

Yes. AI systems can reproduce and sometimes magnify biases already present in research, datasets and human decision-making.

In more detail

AI learns from existing information. If certain populations, languages or study types are underrepresented in training data, outputs may systematically favour dominant perspectives. AI can therefore reinforce existing inequalities unless bias is actively monitored and mitigated.

Why it matters

Bias may affect evidence quality and equity.

Decision rule

Assess whether important groups or evidence sources may be underrepresented.

Common misconception

  • “Larger datasets automatically eliminate bias.”

At a glance

Evidence strength
Strong

Related concepts

Algorithmic Bias Fairness Equity
Key takeaway

More data does not automatically mean less bias.

More on Hallucinations, Bias & Fairness