Evolutionary psychologists have shown that humans' capabilities are quite spiky. Some notable spikes they've discovered:
Thanks for these cool examples! In case I wasn't clear, using a unit circle doesn't mean that the capabilities of the reference species are smooth because whatever you choose as the reference becomes a circle by definition... But then how do we even define smooth, relative to what, computation needed for the task? Seems not useful
Hmm, tricky… maybe average across species, or across AI with different training data nad architectures? I think like "we can do functionally isomorphic logic problems significantly better if they're framed as being about X, rather than about anything else" seem like a genuine spike, compared to some sort of theoretical "neutral" comparison — but it's a lot harder to see how to define that in general.
This type of thinking can be extended to individual humans relative to each other, or to human averages or AI averages. I'd argue that they'd look almost as different as they do between human averages and current AIs.
I'd further argue that this is because humans are actually fully general intelligences; we have strong continual learning, which we can deploy on any topic we're motivated to learn (subject to some sensory and motor limitations). Because we learn whatever we take an interest in, our motivations are the primary determiners of our individual and collective capability shape. And our motivations are in turn heavily shaped by our circumstances, including the culture and society surrounding us.
But that is off of your original topic.
Definitely agree with the first paragraph, it actually briefly crossed my mind as I was writing this up, would be fun to compare the average LWer graph to the average person haha
Mildly agree with the second paragraph I think
Epistemic status: I came up with this an hour ago, this is a fun light-hearted post where I'm enjoying illustrating this idea, which seems true for the most part.
[Crossposted from my blog, BlueprintingHeaven.]
When it comes to the frontier of AI development, we have this concept of "jaggedness". AIs are much better at some kinds of tasks than others.
And to that I say, you are assuming our circle of capabilities is God! How dare you suggest our circle is the Right and True one?
To be clear, jaggedness is a useful term, obviously it's "relative to humans" because we are humans, there are no problems here, I'm sure people are already aware of this and if they're not it's just because it hasn't crossed their minds, I just want to say this fun thing.
https://x.com/LaibaRehmanRJ/status/2107123232675414172
Assume the circle of human capabilities are showing the difficulty level of each task, and let's say we have an infinite number of axes emanating from the centre going from zero difficulty to infinite difficulty, and assume the human capability circle is a unit circle.
Then, if you have a capability level of 1.0 for humans (which is the case on every axis, because it's a unit circle) and 4.0 for an AI, then if you switch to the POV of an AI, such that the unit circle of capability is now for the AI, the human frontier for that specific axis is now 0.25.
There may be some reasons why having a unit circle from the POV of any species doesn't work beyond the problems with using a unit circle to represent human capabilities in the first place. I can't think of them, but I think they probably exist, my mind is vaguely grasping at ideas like that but not well enough to make concrete.
Ooh, it would be cool to see people who build those huge fantasy/sci-fi universes to make capability circles like this for each species! Now that I mention it, could even be fun to do with animals.