> Recent advances in AI, and less recent advances in computing in general showed that a lot of the things we consider “cognitively hard” - from symbolic manipulation to arithmetic, “drawing” and even reasoning - can be reproduced by computer machinery a lot simpler than we originally thought, at a (likely) higher than human level in the near future
Robots still have a hard time navigating terrain and manipulating stuff<p>...<p>> So, here’s the speculation / thesis: exercising, that is, operating our bodies at a higher than baseline level, with everything that entails - energy regulation, unconscious balancing, spatial awareness, etc - requires more computing power than crosswords, sudoku, finding relations between measure theory and first-order logic through category theory, etc<p>I don't buy it. Stuff like "navigating terrain" is something that insects can do often better than robots. It's simply not that hard for animals, which have specialised hardware for this (but on the flipside we lack hardware IEEE compliant floating point units).<p>Exercise is good because the brain is an organ, and general health is good for organs. And because of psychological factors. General and mental health have a huge impact on cognitive performance.<p>Cognitive training is not particularly effective because it's a lot of wistful thinking and snake oil (there might be something to it, but I've seen little evidence that it's particularly effective especially when it comes to transfer).<p>There probably is some cognitive benefit to exercise. Decision making under pressure, discipline, spatial awareness, etc and there would be some transfer to related domains. But I suspect the main effect is just health (which both is better for cognitive performance, and encourages other activities which are good for cognitive performance).