Harvard researchers find AI coding agents shifted the bottleneck to code review
Harvard study: AI coding agents raised code output 30 percent; finished work did not measurably rise, and getting code through review took 49 percent longer.
Read full article →