Minsky, in the 60s, thought that object detection / classification with a camera was worthy of a summer research project for undergrads. Maybe there is a classical algorithmic way to do so (I personally don't believe there is). But I would file that under optimism, since that problem realistically took massive amounts of data (PBs? XBs?) and Machine Learning to get decent at. IN the 60s I doubt there was enough compute in the world to solve that. Which is why I put it under optimism.