General scales unlock AI evaluation with explanatory and predictive power
General scales unlock AI evaluation with explanatory and predictive power Summary This Nature paper introduces a construct-oriented, scalable methodology for evaluating general-purpose AI (tested here on large language models). Instead…
