claim
GPT-4 can sometimes surpass clinicians in estimating disease likelihoods, although both the model and human experts deviate substantially from actual prevalence rates.

Authors

Sources

Referenced by nodes (1)