Scientific Frontline
Deep #convolutional #neural networks (#DCNNs) don’t see objects the way #humans do – using configural shape perception – and that could be dangerous in real-world #AI applications
#ArtificialIntelligence #Neuroscience #sflorg
https://www.sflorg.com/2022/09/ai09172201.html
Even the smartest AI models don’t match human visual processing

York University study highlights how deep-network models take potentially dangerous ‘shortcuts’ in solving complex recognition tasks

Sep 17, 2022 at 9:05pmWeb