Investigations
One claim at a time, traced to source.
Each investigation takes a single, load-bearing claim about AI in global health — a vendor benchmark, a funder statistic, a deployment announcement — and follows it back to the primary evidence, ending in a calibrated verdict you can check. When a number is accurate but incomplete, we say so; we reserve the word “false” for what is actually false.
-
“16% fewer errors”: what the OpenAI–Penda Health study actually measured
OpenAI and Penda Health report that clinicians using an AI copilot in Nairobi made 16% fewer diagnostic errors and 13% fewer treatment errors. The reductions are real, significant and independently physician-rated — but they score the quality of documented decisions, not patient harm, and the one patient-outcome measure came back non-significant. Traced to the primary source.