← Back

Independent International Scientific Panel on Artificial Intelligence

research organizationCredibility: 88%

Why this score? UN body established by General Assembly resolution A/RES/79/325 (26 August 2025); 40 GA-appointed experts co-chaired by Yoshua Bengio and Maria Ressa, publishing consensus scientific assessment with declared evidence gaps and limitations. Its report states it does not represent the views of the United Nations. Held below the top tier because its first report relays some vendor-sourced figures at face value - a '75% of new code' claim traced to a product-launch blog post, and a diagnostic-accuracy figure sourced to trade press relaying a vendor's own test - the same 'AI washing' the report elsewhere warns about.

Tracked Statements (1)

Science cannot yet say with confidence whether the task-level of AI productivity gains will aggregate to economy-wide gains.?

Context: Still unverified, and now sharply illustrated: a registered randomized trial of AI decision support in Kenyan primary care found clinicians' diagnoses and notes significantly improved while patient outcomes did not move (odds ratio 0.77, P = 0.13). Its authors estimate that detecting even a modest patient-level effect would require more than 100,000 patients.