Birdwatch Note Rating
2024-10-02 19:24:48 UTC - NOT_HELPFUL
Rated by Participant: B600BB75350DA1BF79CCD51E2FB34D3FB1499B199E43A8C64529C1CA2767313C
Participant Details
Original Note:
The figures referenced are "diagnostic reasoning scores" not final diagnosis accuracy. Per supplementary eTable 4, GPT-4 had the correct diagnosis in 66% of cases, compared with 62% for the doctors, but this is not a statistically significant difference. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10980135/
All Note Details