The relevant difference is that humans are autopoietic, i.e. humans can reproduce themselves + their infrastructural civilizational substrate, and LLMs aren't. So LLMs depend on humans but not vice versa. "Abundance" of LLM agents is not enough. There is an objective gap in generality determined by the evolutionary function in nature.
Humans are hilariously bad at wilderness survival in the absence of societal knowledge and support. The support doesn't need to be 21st-century-shaped but we do need both physical and social technology to survive and reproduce reliably.
That doesn't matter much, though, because humans live in an environment which contains human civilization. The "holes" in our capabilities don't come up very often.
The right tools could also paper over many of the deficiencies of LLM agents. I don't expect the tools which make groups of LLM agents able to collectively do impressive things to result in particularly human-shaped agents though.
Concretely, sample efficiency is very important if you want a human-like agent that can learn on the job in a reasonable amount of time. It's much less important if you can train once on how to complete each task with a standardized set of tools, and then copy the trained narrow system around as needed.
(Note: perhaps I should say "language-capable agent" rather than "llm-based agent")
I get that individual domesticated humans are not all that general. But there are small groups of people that survive in the wilderness without civilization.
More generally there are objective metrics of success other than autopoiesis. Like, "build a power plant that generates 1 gigawatt". We know humans can do that without LLMs. And we know LLMs can't do that without humans.
Even if there are 100x more LLM agents running around than human agents, that is not going to mean LLMs can build a 1 gigawatt power plant by themselves. Most of the economy could even be organized around LLMs (the way much of the economy is currently organized around computers and the Internet), and that still wouldn't imply LLMs could do the cognitive work required to build a 1 gigawatt power plant. (To make the comparison fair we could say they have access to similar actuators as humans somehow, through robotics)
Concretely, sample efficiency is very important if you want a human-like agent that can learn on the job in a reasonable amount of time. It's much less important if you can train once on how to complete each task with a standardized set of tools, and then copy the trained narrow system around as needed.
This is still a difference in generality though. It seemed earlier you were arguing against an objective difference in generality, as distinct from the organization of the economy.
My point is more that we have millennia of experience building tools and social structures for making humans able to successfully accomplish tasks, and maybe 2 years of experience building tools and structures for making LLM agents able to successfully accomplish tasks.
I do agree that there's some difference in generality, but I expect that if we had spent millennia gathering experience building tools and structures tailored towards making LLMs more effective, the generality failures of LLMs would look a lot less crippling.
If you take a bunch of LLMs and try to get them to collaboratively build a 1GW power plant, they are going to fail mostly in ways like
All of these are failure modes which can be substantially mitigated by better scaffolding of the sort that is hard to design in one shot but easy to iteratively improve over time.
Humans are hilariously bad at wilderness survival in the absence of societal knowledge and support.
what does this mean? it doesn't seem true to me.
Assessments of "general" vs "spiky" capability profiles are secretly assessments of "matches existing infrastructure" vs "doesn't".
Human societies contain human-shaped roles because humans were the only available workers for most of history. Packaging tasks into human-sized, human-shaped jobs was efficient.
Given LLMs, the obvious thing to do is to try to drop them into those roles, giving them the same tools and affordances humans have. When that fails to work, though, we should not immediately conclude that the failure is because LLMs are missing some "core of generality".
When LLM agents become more abundant than humans, as seems likely in the very near term, the most effective shape for a job stops being human-shaped. At that point, we may discover that human capability profiles are the spiky ones.