Even smartest AI models don't match human visual processing: How deep-network models take potentially dangerous 'shortcuts' in solving complex recognition tasks

Open in new window