AI and the mismeasure of science

Before AI, writing papers and grant applications served as costly signals of a scientist’s ability. Today, they can no longer serve that function. The time has come to return to illegible, subjective human evaluations of scientific quality. Rate-limiting authors is a necessary first step.