‘It’s either a panda or a gibbon’: AI winters and the limits of deep learning