A human-accelerated neuron type potentially underlying autism in humans

This makes sense. I do feel like intelligence has hard limits. Maybe knowledge doesn't have limits but intelligence definitely feels like it has limits.

If you consider IQ tests, a lot of it is about seeing a sequence and seeing all possible patterns and then figuring out which one is the most certain/obvious based on the limited sample given. One could imagine all sorts of complex patterns beyond the 'correct answer'. But there's a point it loses all utility value. But it's not right either to assume that the most obvious/simplest pattern is always the right one. Not all logic is elegant, especially not when it comes to human matters.

If you believe in the intelligence -> compression thing, then the hard limit would be the komologorov complexity.

You could imagine that an ever more intelligent person could solve your IQ test pattern puzzle quicker and quicker, or could solve bigger and bigger puzzles of this kind.

Along these lines, I do think that most human problems don't require that much intelligence to solve. There are patterns which are counter-productive and best left unidentified if your goal is to create economic value.

My PoV is that conversations around intelligence should be had alongside conversations about goals.

> My PoV is that conversations around intelligence should be had alongside conversations about goals.

Well, one definition of intelligence is something along the lines of 'better able to achieve your goals (holding resources available constant)'.

It's easy to upgrade the RAM or hard drive on a computer, but it's not possible to upgrade the L2 cache (other than swapping the CPU).