In this post, I will address multiple issues mentioned in the paper "Naturalistic Computational Cognitive Science: Towards generalizable models and theories that capture the full range of natural behavior" by Wilka Carvalho and Andrew Lampinen in their "quest for building general models of natural intelligence" (all quotes are from that article if not specified otherwise).
I think your idea seems likely more correct in concept than other such proposals that you compare.
Whatever is the correct answer, I do believe in principle that the search for efficiency should overlap significantly with intelligence.
Intelligence is extraordinarily efficient and much of what is happening right now in AI seems to be ignoring this principle as we seem to be going with "eventually with enough compute and power it will just happen".
If we are on the right path, we should see less power required and less data required as we improve.
I think your idea seems likely more correct in concept than other such proposals that you compare.
Whatever is the correct answer, I do believe in principle that the search for efficiency should overlap significantly with intelligence.
Intelligence is extraordinarily efficient and much of what is happening right now in AI seems to be ignoring this principle as we seem to be going with "eventually with enough compute and power it will just happen".
If we are on the right path, we should see less power required and less data required as we improve.
Thank you! Consider reading my latest post - https://alexandernaumenko.substack.com/p/opening-open-endedness