Humans are biased towards grokking two-dimensional information arranged in grids because that's how vision works. (Non-V) GPT is not. Its natural gestalt has a vast number of dimensions. It is, and it should, be able to find correlations across data no matter how the input is arranged. (Note that for GPT text is not linear even if it appears so to humans.)
To humans it would be qualitatively more difficult to recognize patterns in four or six or ten-dimensional data than in two dimensions. To GPT, it's just another variation of the same problem, one that's at most quantitatively more difficult.
> And even the valid comparison is only measuring visio-spatial intelligence, not IQ.
You're the first one to mention IQ. The paper definitely does not. In any case "visio-spatial intelligence" is a human concept. GPT doesn't distinguish between "visio-spatial" and any other kind of pattern recognition.
To humans it would be qualitatively more difficult to recognize patterns in four or six or ten-dimensional data than in two dimensions. To GPT, it's just another variation of the same problem, one that's at most quantitatively more difficult.
> And even the valid comparison is only measuring visio-spatial intelligence, not IQ.
You're the first one to mention IQ. The paper definitely does not. In any case "visio-spatial intelligence" is a human concept. GPT doesn't distinguish between "visio-spatial" and any other kind of pattern recognition.