2 ms·
Humans can do infinitely many things because we have general intelligence. Testing whether an AI can play chess or solve Chollet's ARC problems, or some other
by HarHarVeryFunny 2y ago
Humans can do infinitely many things because we have general intelligence.
Testing whether an AI can play chess or solve Chollet's ARC problems, or some other set of narrow skills, doesn't prove generality. If you want to test for generality, then you either have to:
1) Have a huge and very broad test suite, covering as many diverse human-level skills as possible.
and/or,
2) Reductively understand what human intelligence is, and what combination of capabilities it provides, then test for all of those capabilities both individually and in combination.
As Chollet notes, a crucial part of any AGI test is solving novel problems that are not just templated versions (or shallow combinatins) of things the wanna-be AGI has been trained on, so for both of above tests this is key.
- bubblyworld 2y agoI suspect trying to reductively understand intelligence is a bit like trying to reductively understand biology - every level of abstraction is causally influenced by every other level of abstraction, so there just aren't simple primitives you can break everything down into.
- daveguy 2y agoYou can get pretty far with an understanding of biochemical reaction cycles, genetic theory, and protein molecular interactions.
- bubblyworld 2y agoTrying to express the high-level behaviour of an organism in those reductive terms is way beyond science right now, if it's even possible at all. Like have a look at a chart of human metabolic pathways - it's absolute insanity. And those are extremely simplified already!