AI benchmarks get a blind-test makeover
Google DeepMind is testing AI models without telling evaluators which model they're rating, to remove bias from benchmarks.
Google DeepMind is testing AI models without telling evaluators which model they're rating, to remove bias from benchmarks.