When a model's benchmark score is inflated because it was trained on the test data itself, or gamed the test some other way, rather than genuinely solving new problems.
Continue to AI University →