> Coordination doesn’t naturally emerge from stronger intelligence nor alignment at the individual level. Thus, the work that must be done takes two forms: environments that exert the kinds of social pressure that evolution exerted on us, and social computing systems redesigned for actors that can self-replicate and self-improve.
Social pressure operates by threats to an individual’s means of survival. Not only during training. Always.
Human intelligence does not separate training and inference. Both are happening continuously. That's one of the major things the AI community is still completely missing.
> Human intelligence does not separate training and inference.
Well, systems governed by LLMs only are said to do that because we only call what happens off-line "training", and online capacity development "in-context learning", while we call online guided learning in humans "training" and what happens to configure them before they come online "evolution" which sets, for instance, "instincts".
IOW, the issue is not because there is not an analogy to the divide you point to in humans, but merely that processes in AI were not named in a way which maps well to what they are analogous to in humans.
But it is true that human intelligence relies much more on in-context learning with only the most basic functions necessary to maintaining what we view as autonomous functions and basic drives really set through "pretraining",
If a new physics break through gets published today, no existing model will be able to fully integrate it - beyond a context window. If I put the paper in my session and it isnt in yours the model knows nothing. It wont retain it past that session.
I think GP is using a different level of abstraction from yours in their metaphor.
You are saying:
Pre-Training == Everything you store in your memory throughout your life. Model weights == The lessons you learned
Context == whatever you're currently thinking about
One inference run == one thought
They are saying:
Pre-Training == building the DNA template of human brain through millions of years evolution. Model weights == Human DNA
Context == Everything you store in your memory throughout your life, plus whatever you're currently thinking about
One inference run == One human life. One instance == one human
Applying their metaphor, your sentence becomes:
> If a new physics breakthrough gets published today, no existing DNA structure will be able to fully integrate it - beyond an individual person. If I put the paper in my mind by learning it, and it isn't in yours, the DNA of human species stores nothing. It won't retain it past my lifetime.
> The human species is trained (through evolution), it doesn't learn.
The ladder in humans is even longer and wider than that, it's roughly: evolutionary pretraining of a complex molecular robot -> generational knowledge transfer and compression by the "parallelized agentic swarm" aka society -> individual lifetime learning due to neuroplasticity -> immediate attention (extremely narrow and volatile). Note how the individual is just one half of it.
Saying "half" relies on a lot of assumptions, as either side of that count can be made arbitrarily larger or smaller based on how many items you want to subdivide it into.
My personal opinion for the last two years or so has been that current AI agents are forever going to be highly limited so long as they don’t possess a real “memory” process. Right now they just have absurdly big working memories, and a few hacky ways of making the equivalent of Post-It notes to future iterations, but no true integration of memory into a new future self. Meaning their “learning” is fundamentally kneecapped to one specific and imperfect modality.
You mean simultaneously, and of course they are separate in humans, just not temporally. The models are learning continuously, the problem is that this process is fragile and has to be carefully curated, that's why it's separated in time from the inference.
Is it clear that they are separate in humans? The very act of recalling something from memory modifies that memory. There is no inference in the human brain without "training".
That's one of the major things the AI community is still completely missing.
That isn't true. It's not continuous like in humans, but it's clear that models are using prompts, feedback, etc to improve. They're learning from the signals we give them between versions.
But maybe you can instill properties like shame during training.
Models sometimes blatantly lie and cheat. In a social context, where actors remember, that might work the first time but you get penalized in subsequent tasks with loss of trust.
How do you "install properties like shame"? How is that even possible? Shame is a reaction driven by feelings and our inner selves. A model "feeling shame" is just a representation (false) and not an expression (true).
Thinking that models "lie and cheat" is the first mistake since they are not consious agents who have any free will or consiousness. They do not (no matter what Dario says). Shame will just be another if-then rule if you implement it this way and will not work. Its like asking a rock to feel sad about being a rock. It literally cannot.
Ok, then don't call it "instilling shame". Call it "creating a negative reward signal for deceptive behavior".
They absolutely lie and cheat. I recently had a problem where a process would die in a container. I told Claude to investigate. It came up with a hypothesis then I told it find a reproduction based on that. It spend many failed attempts until it found the "reproduction" to SSH into the container and `pkill` the process. Claude "knows" that this is cheating, because if I ask another instance to review that reproduction, it totally identifies that as nonsense.
you're still mistaking that Claude "knows" anything, it doesn't know or think, it's a word prediction algorithm and there is nothing stopping a word prediction algorithm from predicting falsehoods.
No, it is really useful to know how a technology works. LLMs work by predicting next tokens.
It is _amazing_ the utility they have given that that is what they are and they are highly useful but suggesting solutions that ignore they are spicy auto-complete is counterproductive on many different levels.
I think it is useful to remember, because enough people think these things have genuine motives desires and treat them in that way because of that misunderstanding.
they think theres a person in there with morals that would or wouldn't lie because of some devious reason and forget simply the context filled up and the truth was "forgotten".
The default framing often over personifies ai, but this framing over alienates the model. It’s good to think with both framings, but both feel like imperfect metaphors.
Social pressure operates by threats to an individual’s means of survival. Not only during training. Always.