Humans do something very well that machines cannot, and that’s transfer learning. This is how we take what we know in one field and apply it to another. You can show a human a photograph of an item, just train it with one piece of data, and it can recognize that in all different contexts and all different situations.
Computers have a hard time with this. Why is it hard? Because, generally speaking, they don’t know what to transfer. Imagine a fish swimming in the ocean, and a specimen of the same species in a jar of formaldehyde in a laboratory. What do they have in common, and what’s different? Are they the same temperature? No. Are they the same weight? Yes. Are they the same color? Maybe. Do they react the same way when poked? No. When dropped from a tall building? Yes. Do they smell the same? No. Do they have the same number of bones? Yes. How do we know this?
This exercise is effortless for a human, and we aren’t really sure that we understand how we do it. Of course, we know generally that we have a lifetime of history with similar items, but the way we make associations about what knowledge transfers to what situation is still a mystery.