> Applied statistics is a
far more precise descriptor, “but no one wants to use that term, because
it’s not as sexy.”
This really hit me some time back when I was explaining AI to a friend. After about 10 mins of rambling about LLMs and mentioning the attention paper like I knew what I was talking about, it ended with “oh so it’s just a really advanced auto correct”
As an MLE I feel these takes are too reductionist.
You could say the (nearly) same thing about search. And content recommendation. And clustering. And topic modeling. And outlier detection. And spam filtering. And image diffusion. And dimension reduction. And...
There's a lot in common between these things, but there's also a lot cool and different!
For transformers in particular, it's pretty cool that you get some WILD emergent properties simply from scaling up.
So yes, it's just a next token predictor, but I'm just a bundle of nerves and meat. I don't get a lot out of those descriptions.
Some concrete facts about LLMs are explained by their next token predictor nature. Every time it says "wait, that's wrong." instead of generating the correct thing the first time.
I think that's relatively emergent too though! BERT never really did that (at least to my recollection), presumably because its training was never sufficient for it to develop corrective reasoning in a chain of thought.
I don't know what the parent poster means, but if you look at how LLMs are trained, and how it's trained on human communication, you should find it interesting that humans often correct themselves. They also often create facts out of nowhere, and do this crazy thing called dreaming.
If you're an LLM trying to symbolically emulate human behavior, especially when coding, you're going to see git commits, where people write a bunch of code and then go, oh wait, I should fix that!
From the perspective of an outside observer, making mistakes, and then correcting the mistakes, is how you code.
In fact, an outside observer might think that this is how you teach people to code. You make a mistake, then you say, oh, here's the mistake I made, and here's why, and then you fix it.
So I think true emergent behavior will be demonstrated, when an LLM says "Hey wait a minute! They're actually making real mistakes, and they're actually correcting, it's them not on purpose?!", and then stops hallucinating, and stops making mistakes as it codes.
The same is true of bug trackers. You submit a bug request and you often have a patch. Then people discuss it. From a MLM's perspective, you're supposed to write poor code, or code that's not perfect, and then have a conversation.
If I had the resources to train a large-scale LLM, I would clone, for example, GitHub. I would then remove everything where people are fixing broken code.
The outcome would be very interesting.
I've been looking at ways to make enhanced long-term memory stores for LLMs, and there's lots of problems with shifting symbolic relationships if you do it wrong, but definitely once there's true long-term memory, and adaptive behavio,r I think that's the only way you're going to get true real emergent behavior.
I mean sure, but BERT can be and often is used as a next token predictor / generator.
I could have used any series of NLP examples, the point is this CoT behavior only emerges when you get to a certain scale (and training style, presumably).
It's a little different than that. Your bundle of nerves and meat is not static. It changes over time.
To me the heart of the "next token predictor" is that the distributions are static. You can manipulate what you feed into it through context (and a lot of interesting engineering has been applied there through CoT and other techniques to manipulate the prompt). But these models as implemented will never be able try things and learn from mistakes or adapt. They are a set of weights frozen in time. A set of distributions derived from the original data that created them.
That's not quite true though. The fact that most models are in practice non-deterministic has been a huge point of contention.
And there's nothing inherently stopping labs from continuously fine-tuning the weights after every new invocation. It's just a difficult (not to mention expensive) software and ML problem.
How are they non-deterministic? Ok there is random involved but for watermarking results the random is substituted by pseudo-random, otherwise it won’t work. If you control the randomness, you should be able to reproduce answers to prompts in equal context 100%.
I was not saying that they are deterministic, rather that the distributions (aka weights) are fixed. A model as deployed today at anthropic/open ai/etc is not learning beyond the context as far as I know.
What prevents continuous fine-tuning from what I understand is catastrophic forgetting. You can do things like RLHF which are built to minimize the damage but that is more about bringing out capabilities of the base model than incorporating new knowledge (at least from my understanding, I am obviously not a researcher at a lab).
Yes, catastrophic forgetting is absolutely one of the problems that needs to be solved to enable something like this.
My broader point is just that there's nothing inherent to the structure of LLMs that stops them from updating their weights and continuously learning from environmental feedback in the way humans do, and there's already solid templates for how they could push even further in that direction.
But as an assessment of the current state, I agree with you, LLMs lag humans severely in ability to self-update.
3. Ability of AGENT-HARNESSES to learn
with the help of the human user.
So agents learn, LLM already knows
everything it will ever know, and
ESPECIALLY it has already learned
how to understand human language.
No 3. above means there is no danger
of the LLM getting corrupted. But
the agents running on user's machine
learn on behalf of that user who
shares the machine with them.
>My broader point is just that there's nothing inherent to the structure of LLMs that stops them from updating their weights and continuously learning from environmental feedback in the way humans do, and there's already solid templates for how they could push even further in that direction.
"LLM" is a branded model as a product. Of course it could be anything, as long as it fulfills the product category.
But we live in reality, we can only look at what models are out there and we see that they don't do any of those things and yet we're supposed to act as if these models already do.
It's not "not quite true", it's literally true because alternative architectures like RNNs and Mamba fully update their own internal states, whereas transformers only append to the context.
RNNs and Mamaba do not update their weights, but you could hypothetically scale the internal state to be as big as Fable's and GPT 6's parameters.
It wouldn't require perfect accuracy, just rough accuracy and a human to confirm, and it's already more than good enough for that. I do not understand this confusion surrounding modern math.
Essentially all of economic theory is aimed at explaining, not predicting. The distinction between the two goals [0] is sometimes under-appreciated within the profession, and almost always under-appreciated outside of it.
Most predictive tools in economics and finance have “surprisingly” little economic content; but once you understand the distinction between the two goals, it should be unsurprising that predictive models tend to make few economic assumptions, relying rather on general statistical techiniques or on econometrics that incorporate a minimum of theory [1]. From that understanding comes the humbling realization that predicting the future is quite difficult in a context in which the relevant processes are continually seeking an equilibrium that often implies unpredictability. [2]
I’m not an economist, but I do a lot of applied financial-economic modeling. State-of-the-art LLMs are really, really terrible at economic intuition. They will hinder, not help, in formulating an economic model, which is a process of coming up with a set of modeling assumptions that lead to a useful (implicitly, tractable) model. LLMs are, however, quite good at math, and I’ve found them very useful in iterating through different sets of modeling assumptions to identify those that lead somewhere useful. Not having to work out all of the mathematical details myself, and thereby avoiding getting lost in the weeds and being better able to maintain a higher-level perspective on what I’m trying to accomplish, has accelerated my work immensely. But it’s a process of leading the LLM by the nose the whole time and asking it to fill in the details.
I should note, thought, that if you indotend “AI” to mean more than LLMs, them yes, there is starting to be a lot of good work done on predictive economic models that use specialized neural networks as black-box functions to compute model quantities that are otherwise difficult to come up with, just as is also happening in applied physics and other fields.
1. Many explanatory economic models refer to quantities that are fundamentally or practically unobservable or unidentifiable. Much of economics is built on models that were designed to provide a formal, logical basis for understanding the economic world, which is often quite unintuitive. (For example, many intelligent people uneducated in economics exhibit intuitions opposite of basic economic ideas like opportunity cost or comparative advantage.) Models of this sort have been very influential in determining the trajectory of economic thought, but they are often effectively impossible to calibrate to the real world.
2. The most influential and effective economic ideas fall into a third class: ideas that have created their own reality by shaping the way people think in a way that gives rise to the results the models explain or predict. This phenomenon is most evident in finance, where ideas like the various forms of the efficient market hypothesis, the CAPM, and the Black–Scholes model and its follow-one have arguably provided a framework that has reshaped the ways financial practitioners behave to such an extent that financial markets now conform much more closely to what the models describe than was formerly the case. Donald MacKenzie’s book An Engine, Not a Camera is an excellent study of this phenomenon: https://mitpress.mit.edu/9780262633673/an-engine-not-a-camer...
>Much of economics is built on models that were designed to provide a formal, logical basis for understanding the economic world, which is often quite unintuitive.
I disagree. It's purposefully unintuitive.
>(For example, many intelligent people uneducated in economics exhibit intuitions opposite of basic economic ideas like opportunity cost or comparative advantage.)
Most people don't believe in comparative advantage. They believe in something that economists can explain away as comparative advantage.
All unconsumed fixed size investments will result in something that is mathematically the same as comparative advantage. This is the intuitive view that people have. You go to university and get a 5 year degree. Now your cost basis for work that suits your expertise is much lower than for work that is out of expertise. A worker buys an expensive machine, now the cost basis for hiring the guy with the machine is lower than buying your own machine.
This also explains why specialization emerges: All specialization is basically a form of an investment that has some residual left over results that can be monetized in the future. If there was no residual it would be as if you forgot your education and at that point the investment is fully consumed and you turn back into a non-specialized worker.
All of this is incredibly intuitive, but economists instead insist on an invisible "factor" [0] to drive efficient production.
[0] The "factor" concept implies comparative advantage exists first rather than emerges as a result of past decisions.
Here's the thing about economists... The loudest ones don't want to be correct, they want to be influential. The ones who can actually make good predictions work for banks and hedge funds lol.
- Many more mediocre papers written (mediocre ideas, implementation, claude-isms everywhere)
- Much easier to try every possible combination of a regression in order to show the result you want (same for theorists).
The one thing I'm happy about is it's now much easier to extract historical data from old documents from Google Books. Still not perfect, but takes you 95% there. And creating plots and datavis just for quick exploration is super fast.
> Applied statistics is a far more precise descriptor, “but no one wants to use that term, because it’s not as sexy.”
This really hit me some time back when I was explaining AI to a friend. After about 10 mins of rambling about LLMs and mentioning the attention paper like I knew what I was talking about, it ended with “oh so it’s just a really advanced auto correct”
As an MLE I feel these takes are too reductionist.
You could say the (nearly) same thing about search. And content recommendation. And clustering. And topic modeling. And outlier detection. And spam filtering. And image diffusion. And dimension reduction. And...
There's a lot in common between these things, but there's also a lot cool and different!
For transformers in particular, it's pretty cool that you get some WILD emergent properties simply from scaling up.
So yes, it's just a next token predictor, but I'm just a bundle of nerves and meat. I don't get a lot out of those descriptions.
Obligatory link to the classic copy-pasta:
> They're Made out of Meat
https://web.mit.edu/people/dpolicar/writing/prose/text/think...
Some concrete facts about LLMs are explained by their next token predictor nature. Every time it says "wait, that's wrong." instead of generating the correct thing the first time.
I think that's relatively emergent too though! BERT never really did that (at least to my recollection), presumably because its training was never sufficient for it to develop corrective reasoning in a chain of thought.
What exactly do you mean with "emergent" here?
I don't know what the parent poster means, but if you look at how LLMs are trained, and how it's trained on human communication, you should find it interesting that humans often correct themselves. They also often create facts out of nowhere, and do this crazy thing called dreaming.
If you're an LLM trying to symbolically emulate human behavior, especially when coding, you're going to see git commits, where people write a bunch of code and then go, oh wait, I should fix that!
From the perspective of an outside observer, making mistakes, and then correcting the mistakes, is how you code.
In fact, an outside observer might think that this is how you teach people to code. You make a mistake, then you say, oh, here's the mistake I made, and here's why, and then you fix it.
So I think true emergent behavior will be demonstrated, when an LLM says "Hey wait a minute! They're actually making real mistakes, and they're actually correcting, it's them not on purpose?!", and then stops hallucinating, and stops making mistakes as it codes.
The same is true of bug trackers. You submit a bug request and you often have a patch. Then people discuss it. From a MLM's perspective, you're supposed to write poor code, or code that's not perfect, and then have a conversation.
If I had the resources to train a large-scale LLM, I would clone, for example, GitHub. I would then remove everything where people are fixing broken code.
The outcome would be very interesting.
I've been looking at ways to make enhanced long-term memory stores for LLMs, and there's lots of problems with shifting symbolic relationships if you do it wrong, but definitely once there's true long-term memory, and adaptive behavio,r I think that's the only way you're going to get true real emergent behavior.
BERT isn't a next token predictor. It predicts a single token based on the whole surrounding context in both directions.
I mean sure, but BERT can be and often is used as a next token predictor / generator.
I could have used any series of NLP examples, the point is this CoT behavior only emerges when you get to a certain scale (and training style, presumably).
It's a little different than that. Your bundle of nerves and meat is not static. It changes over time.
To me the heart of the "next token predictor" is that the distributions are static. You can manipulate what you feed into it through context (and a lot of interesting engineering has been applied there through CoT and other techniques to manipulate the prompt). But these models as implemented will never be able try things and learn from mistakes or adapt. They are a set of weights frozen in time. A set of distributions derived from the original data that created them.
That's not quite true though. The fact that most models are in practice non-deterministic has been a huge point of contention.
And there's nothing inherently stopping labs from continuously fine-tuning the weights after every new invocation. It's just a difficult (not to mention expensive) software and ML problem.
How are they non-deterministic? Ok there is random involved but for watermarking results the random is substituted by pseudo-random, otherwise it won’t work. If you control the randomness, you should be able to reproduce answers to prompts in equal context 100%.
I was not saying that they are deterministic, rather that the distributions (aka weights) are fixed. A model as deployed today at anthropic/open ai/etc is not learning beyond the context as far as I know.
What prevents continuous fine-tuning from what I understand is catastrophic forgetting. You can do things like RLHF which are built to minimize the damage but that is more about bringing out capabilities of the base model than incorporating new knowledge (at least from my understanding, I am obviously not a researcher at a lab).
Yes, catastrophic forgetting is absolutely one of the problems that needs to be solved to enable something like this.
My broader point is just that there's nothing inherent to the structure of LLMs that stops them from updating their weights and continuously learning from environmental feedback in the way humans do, and there's already solid templates for how they could push even further in that direction.
But as an assessment of the current state, I agree with you, LLMs lag humans severely in ability to self-update.
LLMs are "readonly" I guess for several reasons:
1. Technical cost of updating the mode.
2. Inability to trust every user's "truth".
3. Ability of AGENT-HARNESSES to learn with the help of the human user.
So agents learn, LLM already knows everything it will ever know, and ESPECIALLY it has already learned how to understand human language.
No 3. above means there is no danger of the LLM getting corrupted. But the agents running on user's machine learn on behalf of that user who shares the machine with them.
>My broader point is just that there's nothing inherent to the structure of LLMs that stops them from updating their weights and continuously learning from environmental feedback in the way humans do, and there's already solid templates for how they could push even further in that direction.
"LLM" is a branded model as a product. Of course it could be anything, as long as it fulfills the product category.
But we live in reality, we can only look at what models are out there and we see that they don't do any of those things and yet we're supposed to act as if these models already do.
It's not "not quite true", it's literally true because alternative architectures like RNNs and Mamba fully update their own internal states, whereas transformers only append to the context.
RNNs and Mamaba do not update their weights, but you could hypothetically scale the internal state to be as big as Fable's and GPT 6's parameters.
LLMs are applied statistics in the exact same way that you are applied statistics.
Thanks for the book recommendation.
Didn't know that MIT press features open access books.
Weird title, considering the seminal Yudkowsky paper…
Which paper are you referring to?
Speaking how which, how are economists using AI? Are they getting better at making predictions?
The added speed of AI tools means they're now able to predict 18 of the next 10 recessions.
That would require AI to be accurate
It wouldn't require perfect accuracy, just rough accuracy and a human to confirm, and it's already more than good enough for that. I do not understand this confusion surrounding modern math.
Essentially all of economic theory is aimed at explaining, not predicting. The distinction between the two goals [0] is sometimes under-appreciated within the profession, and almost always under-appreciated outside of it.
Most predictive tools in economics and finance have “surprisingly” little economic content; but once you understand the distinction between the two goals, it should be unsurprising that predictive models tend to make few economic assumptions, relying rather on general statistical techiniques or on econometrics that incorporate a minimum of theory [1]. From that understanding comes the humbling realization that predicting the future is quite difficult in a context in which the relevant processes are continually seeking an equilibrium that often implies unpredictability. [2]
I’m not an economist, but I do a lot of applied financial-economic modeling. State-of-the-art LLMs are really, really terrible at economic intuition. They will hinder, not help, in formulating an economic model, which is a process of coming up with a set of modeling assumptions that lead to a useful (implicitly, tractable) model. LLMs are, however, quite good at math, and I’ve found them very useful in iterating through different sets of modeling assumptions to identify those that lead somewhere useful. Not having to work out all of the mathematical details myself, and thereby avoiding getting lost in the weeds and being better able to maintain a higher-level perspective on what I’m trying to accomplish, has accelerated my work immensely. But it’s a process of leading the LLM by the nose the whole time and asking it to fill in the details.
I should note, thought, that if you indotend “AI” to mean more than LLMs, them yes, there is starting to be a lot of good work done on predictive economic models that use specialized neural networks as black-box functions to compute model quantities that are otherwise difficult to come up with, just as is also happening in applied physics and other fields.
0. https://www.stat.berkeley.edu/~aldous/157/Papers/shmueli.pdf
1. Many explanatory economic models refer to quantities that are fundamentally or practically unobservable or unidentifiable. Much of economics is built on models that were designed to provide a formal, logical basis for understanding the economic world, which is often quite unintuitive. (For example, many intelligent people uneducated in economics exhibit intuitions opposite of basic economic ideas like opportunity cost or comparative advantage.) Models of this sort have been very influential in determining the trajectory of economic thought, but they are often effectively impossible to calibrate to the real world.
2. The most influential and effective economic ideas fall into a third class: ideas that have created their own reality by shaping the way people think in a way that gives rise to the results the models explain or predict. This phenomenon is most evident in finance, where ideas like the various forms of the efficient market hypothesis, the CAPM, and the Black–Scholes model and its follow-one have arguably provided a framework that has reshaped the ways financial practitioners behave to such an extent that financial markets now conform much more closely to what the models describe than was formerly the case. Donald MacKenzie’s book An Engine, Not a Camera is an excellent study of this phenomenon: https://mitpress.mit.edu/9780262633673/an-engine-not-a-camer...
>Much of economics is built on models that were designed to provide a formal, logical basis for understanding the economic world, which is often quite unintuitive.
I disagree. It's purposefully unintuitive.
>(For example, many intelligent people uneducated in economics exhibit intuitions opposite of basic economic ideas like opportunity cost or comparative advantage.)
Most people don't believe in comparative advantage. They believe in something that economists can explain away as comparative advantage.
All unconsumed fixed size investments will result in something that is mathematically the same as comparative advantage. This is the intuitive view that people have. You go to university and get a 5 year degree. Now your cost basis for work that suits your expertise is much lower than for work that is out of expertise. A worker buys an expensive machine, now the cost basis for hiring the guy with the machine is lower than buying your own machine.
This also explains why specialization emerges: All specialization is basically a form of an investment that has some residual left over results that can be monetized in the future. If there was no residual it would be as if you forgot your education and at that point the investment is fully consumed and you turn back into a non-specialized worker.
All of this is incredibly intuitive, but economists instead insist on an invisible "factor" [0] to drive efficient production.
[0] The "factor" concept implies comparative advantage exists first rather than emerges as a result of past decisions.
Not sure comparative advantage needs past decisions to happen - could just be randomly distributed resources, for example.
Here's the thing about economists... The loudest ones don't want to be correct, they want to be influential. The ones who can actually make good predictions work for banks and hedge funds lol.
That would require economists to abandon perfect rationality and perfect information so no.
Honestly, it's a bit of a disappointment
- Many more mediocre papers written (mediocre ideas, implementation, claude-isms everywhere)
- Much easier to try every possible combination of a regression in order to show the result you want (same for theorists).
The one thing I'm happy about is it's now much easier to extract historical data from old documents from Google Books. Still not perfect, but takes you 95% there. And creating plots and datavis just for quick exploration is super fast.