Darpa Wants to Solve Science’s Reproducibility Crisis With AI

Social science has an image problem—too many findings don’t hold up. A new project will crank through 30,000 studies to try to identify red flags.