WHO dossier/Risks & Challenges
Part 06 · Risks & challenges
AI can scale weak evidence as quickly as strong evidence.
WHO highlights risks across opacity, accountability, bias, discrimination, equity, privacy, cybersecurity, resource intensity, regulatory gaps, and erosion of human expertise.
Risk map
What needs a control.
policy riskOutputs become hard to challenge
Policy users may not see how sources were selected, weighted, summarized, or transformed.
Missing voices become missing evidence
Bias can enter through data, model design, evidence availability, language, geography, and power.
Human expertise can be displaced
Automation pressure can reduce contestability, judgement, and accountable deliberation.
Reflection
Risk questions for teams.
review logicWhat can the AI workflow miss? Who could be harmed by a misleading output? Which source claims are uncertain? What must be reviewed before use? What signal would make the team pause or revise the policy output?