Posts
Comments
That’s a fantastic memory aid for this concept, much appreciated! Crafting games in general give ample examples to internalize this kind of bootstrap mentality. Also for quickly scaling to the next anvil-equivalent. As you touched upon, real life has a deep crafting tree, with anvil problems upon anvil problems. Something that took me far too long to learn, if you got your anvil, but still don't find yourself were you want to be, it pays to find the next anvil problem quickly. If you still have a lot of distance to cover, don't get bogged down by things that won't get you the next anvil-equivalent.
In a certain way, relationships have their own anvils. There are thresholds of trust, communication modes, that take investment. However, they also unlock completely new options, particularly when addressing challenges or navigating high-stress situations. I sometime notice, in me and others, a neglect to do serious work on relationships during good times, then lacking the tools to handle difficulties when they arise.
Hey, thanks for taking the time to answer!
First, I want to make clear that I don’t believe LLMs to be just stochastic parrots, nor do I doubt that they are capable of world modeling. And you are right to request some more specifically stated beliefs and predictions. In this comment, I attempted to improve on this, with limited success.
There are two main pillars in my world model that make me, even in light of the massive gains in capabilities we have seen in the last seven year, still skeptical of transformer architecture scaling straight to AGI.
- Compute overhangs and algorithmic overhangs are regularly talked about. My belief is that a data overhang played a significant role in the success of transformer architecture.
- Humans are eager to find meaning and tend to project their own thoughts onto external sources. We even go so far as to attribute consciousness and intelligence to inanimate objects, as seen in animistic traditions. In the case of LLMs this behaviour could lead to an overly optimistic extrapolation of capabilities from toy problems.
On the first point:
My model of the world circa 2017 looks like this. There's a massive data overhang, which in a certain sense took humanity all of history to create. A special kind of data, refined over many human generations of "thinking work", crystalized intelligence. But also with distinct blind spots. Some things are hard to capture with the available media, others we just didn't much care to document.
Then transformer architecture comes around, is uniquely suited to extract the insights embedded in this data. Maybe better than the brains that created it in the first place. At the very least it scales in a way that brains can't. More compute makes more of this data overhang accessible, leading to massive capability gains from model to model.
But in 2024 the overhang has been all but consumed. Humans continue to produce more data, at an unprecedented rate, but still nowhere near enough to keep up with the demand.
On the second point:
Taking the globe representation as an example, it is unclear to me how much of the resulting globe (or atlas) is actually the result of choices the authors made. The decision to map distance vectors in two or three dimensions seems to change the resulting representation. So, to what extent are these representations embedded in the model itself versus originating from the author’s mind? I'm reminded of similar problems in the research of animal intelligence.
Again, it is clear there’s some kind of world model in the LLM, but less so how much this kind of research predicts about its potential (lack of) shortcomings.
However, this is still all rather vague; let me try to formulate some predictions which could plausibly be checked in the next year or so.
Predictions:
- The world models of LLMs are impoverished in weird ways compared to humans, due to blind spots in the training data. An example would be tactile sensations, which seem to play an important role in the intuitive modeling of physics for humans. Solving some of the blind spots is critical for further capability gains.
- To elicit further capability gains, it will become necessary to turn to data which is less well-suited for transformer architecture. This will lead to escalating compute requirements, the effects of which will already become apparent in 2025.
- As a result, there will be even stronger incentives for:
- Combining different ML architectures, including transformers, and classical software into compound systems. We currently call this scaffolding, but transformers will become less prominent in these. “LLMs plus some scaffolding” will not be an accurate description of the systems that solve the next batch of hard problems.
- Developing completely new architecture, with a certain chance of another "Attention Is All You Need", a new approach gaining the kind of eminence that transformers currently have. The likelihood and necessity of this is obviously a crux, currently I lean towards a. being sufficient for AGI even in the absence of another groundbreaking discovery.
- Automated original ML research will turn out to be one of the hard problems that require 3.a or b. Transformer architecture will not create its own scaffolding or successor.
Now, your comment prompted me to look more deeply into the current state of machine learning in robotics and the success of decision transformers and even more so behaviour transformers disagree with my predictions.
Examples:
https://arxiv.org/abs/2206.11251
https://sjlee.cc/vq-bet/
https://youtu.be/5_G6o_H3HeE?si=JOsTGvQ17ZfdIdAJ
Compound systems, yes. But clearly transformers have an outsized impact on the results, and they handled data which I would have filed under “not well-suited” just fine. For now, I’ll stick with my predictions, if only for the sake of accountability. But evidently it’s time for some more reading.
The performance of o1 in the first linked paper is indeed impressive, especially on what they call mystery blocksworld. Would not have expected this level of improvement. Do you know of any material that goes into more detail on the RL pre-training of o1?
I do take issue with the conclusion that reasoning in the confines of toy problems is sufficient to scale directly to AGI, though. The disagreement might stem from differing definitions of AGI. LLMs (or LRMs) exist in an environment provided by humans, including the means to translate LLM output into "action". LLMs continue to be very sensitive to the properties of this environment. Humans created the languages that LLMs use to reason, as abstraction layers on top of our perception of reality, to help us communicate, reason and to control machines we build, under the constraint of a limited bandwidth. Everything LLMs do is routed through human understanding of reality. And is for the most part still limited by what we care to make accessible via language. Humans routinely deal with a degree of complexity in our environment which isn't even on the map we provided LLMs with. There is no conveniently pre-processed corpus of data to allow for LLM architecture to bridge this gap. Synthetic data only gets you so far.
You acknowledge the importance of scaffolding, but I think the term understates just how much of the actual intelligence of AGI could end up "in there". It might well turn out to be a far harder problem than language-bound reasoning. You seem to have a different view and I'd be very interested in what underpins your conclusions. If you have written more on the topic or could point me in the direction of what most strongly influenced your world model in this regard, I'd love to read it.
I'd put a reasonably high probability (5%) on orcas and several other species having all the necessary raw mental capacity to be "uplifted" in just a few (<20) generations with technology (in the wider sense) that has been available for a long time. Being uplifted means here the ability to intellectually engage with us on a near-equal or even equal footing, to create culture, to actively shape their destiny. Humans have been training, selecting, shaping other animals since before the dawn of history. Whenever we did so, it was with the goal of improving their use as tools or resources. Never, to my knowledge, has there been a sustained effort to put these abilities to use for the sole purpose of the mental and cultural flourishing of another species. It is my belief that many other universal learning machines beside the human brain have been produced by evolution, but just lack or lacked the right training environment for the kind of run-away development the homo genus went through, for various reasons.
Could "uplifted" orcas outperform humans on hard scientific problems? Would they care to? I don't know, but I'd love to find out.
Indeed, I would have very much preferred to see other animal minds elevated before we turned to the creation of artificial ones. To explore a wider space of minds and values, to learn more about what an intelligent species can be, before we believed ourselves ready to "build" intelligence from scratch. But it seems at least half a century too late for this now.
Then I misunderstood your original comment, sorry. As a different commenter wrote, the obvious solution would be to only engage with interesting people. But, of course, unworkable in practice. And "social grooming" nearly always involves some level of talking. A curse of our language abilities, I guess. Other social animals don't have that particular problem.
The next best solution would be higher efficiency, more socializing bang for your word count buck, so to speak. Shorter conversations for the same social effect. Not usually a focus of anything billed as conversation guide, for obvious reasons. But there are some methods aimed at different goals that, in my experience, also help with this as a side effect.
I understand, for someone with a strong drive to solve hard problems, there's an urge for conversations to serve a function, exchange information with your interlocutor so things can get done. There's much to do and communication is already painfully inefficient at it's best.
The thing is, I don't think the free-association game is inefficient, if one is skilled at it. It's also not all that free. The reason it is something humans "developed" is because it is the most efficient way to exchange rough but extensive models of our minds with others via natural language. It acts a bit like a ray tracer, you shoot conversational rays and by how they bounce around in mental structures, the thought patterns, values and biases of the conversation partners are revealed to each other. Shapes become apparent. Sometimes rays bounce off into empty space, then you need to restart the conversation, shoot a new ray. And getting better at this game, keeping the conversation going, exploring a wider range of topics more quickly, means building a faster ray tracer, means it takes less time to know if your interlocutor thinks in a way and about topics which you find enlightening/aesthetically pleasing/concretely useful/whatever you value.
Or to use a different metaphor, starting with a depth-first search and never running a breadth-first search will lead to many false negatives. There are many minds out there that can help you in ways you won't know in advance.
So if the hard problems you are working on could profit from more minds, it pays off to get better as this. Even if it has not much intrinsic value for you, it has instrumental value.
Hope this doesn't come across as patronizing, definitely not meant that way.