This is nonsense. These nets regurgitate the most average possible output given inputs. By definition. You’re conflating and hyping and mixing up so many weird things at once.

You literally couldn’t get GPT to come up with a single novelty if you tried. It’s all remixing existing content, and again, doing so in a way to fit the average of the dataset.

When you realize this you realize it has no intelligence as we most typically define it (novel solutions to novel problems).. its not AI. Call it what it is: a beautifully advanced way to regurgitate the exact most popular (mundane) reply you’d expect given a huge dataset.

It’s sort of good for studying what already exists. It won’t even really ever show you the edges though so it’s actually almost dangerously deceptive as evidenced by this absurd rounding up people are doing. If you want to learn the gist of anything, ask GPT. If you want to know anything in depth, GPT in fact will only mislead you towards genericity, platitudinous mediocrity.

Has GPT ever told you "let me think" and then you notice it actually pause for a few seconds? No. Its output is fixed cost. GPT can't think in depth because it's not allowed to. This is something I already covered above by noting that current NN models are a DAG. We're already experimenting with variations that can iterate internally, and therefore think deeper, using still their existing knowledge. And the results change dramatically.

You're also not right that GPT can't produce anything novel. Here's a tiny, modest example:

https://sumplete.com/about/

I won't waste my time, as I see you didn't read what you responded to the first time. But you need to start paying attention. You denial & empty confidence in what's happening will only get you so far.

> We're already experimenting with variations that can iterate internally, and therefore think deeper, using still their existing knowledge. And the results change dramatically.

This is still just deterministic except for a bit of noise to vary the output. There is no "thinking" going on. What you get is still just the most statistically likely output based on the training data.

You're also deterministic. Except for a bit of noise. This is less of a slam dunk argument than many think it is.

Considering my former co founder works at OpenAI and we spent many nights working on and talking about related ML with and around some of the best people in the world, and considering I’ve implemented language models multiple times myself, please try and be more specific on where I’m wrong as I know many people who work in ML share this concern about people thinking it’s more than it is.

It’s very useful no doubt, but your comment is pure extrapolation from maladjusted priors.

Your example proves my point.

Your goalpost moved from "nothing novel" to "it's novel, but it's not interesting to me". I'll let not the years ahead, but literally the DAYS ahead prove you wrong. Because things are changing quite rapidly lately.

I've seen ChatGPT draw logical conclusions from (most-likely) previously unseen axioms. I've seen it invent ideas that (most-likely) were not in his training set. I've seen it fail too, but when it succeeds I can not substantially distinguish it from the novelty or logical reasoning of a human.

On niche subjects, it fails often. But you can get it to correct itself by making it think logically by spelling out axioms, which seems to fit your definition of novelty.

All the arguments I've read about it not being intelligent seem to stem from some form of essentialism. Right now it's dumb intelligent but I think giving it an adjustable recursive depth, similarly to what BulgarianIdiot suggests, will go a long way to improving it given its base capabilities.

Indeed. And telling it to spell out the steps in math solution is essentially (ab)using the chat log as the memory for an iteration loop.

This demonstrates then when you instruct it "break down the problem and solve it iteratively" it comes out with superior solutions. More accurate, more novel. Ergo, then if it was allowed to iterate in this way internally, it'd produce better answers by default.

This thread sums up the same idea: https://twitter.com/ItIsFinch/status/1634730897520951296

Yeah, I think it's going to be a promising area of research. It's known as Chain-Of-Thought (CoT) in the literature and there are a few papers already out:

https://openreview.net/forum?id=_VjQlMeSB_J

https://openreview.net/forum?id=5NTt8GFjUHkr

https://openreview.net/forum?id=qFVVBzXxR2V

I'm also looking forward to explicit long-term memory mechanisms that don't rely on the main architecture's weights nor on CoT.

There's no memory or iteration log. You are just priming the (more or less deterministic even...) predictor with some history. You are doing a lot of work yourself, setting a factually correct context (and this has severe limits...).

As for "we just need a way to include iterations": isn't this what attention is supposed to do (kind of dynamically updating weights). The usual way to really update weights are variations on gradient descent. Can you link a paper outlining how we integrate your proposal into the current framework (which took around 20years to mature)? Otherwise your statements are Sci-Fi (at the flying cars-level).

Oh so there's no memory, just recallable history.

Oh so it's not learning by example, it's just primed by a sample.

Oh so there's no iteration of thought, just a loop of prediction.

OK.

Also, the papers you asked about: https://news.ycombinator.com/item?id=35115563

BTW, people are already implementing recursive/iterative queries on the ChatGPT API, and getting promising results.

The langchain project is an example of the iterative queries approach. It comes with constructs for working memory, factual lookup / calculation agents, etc.

https://github.com/hwchase17/langchain

The general (non-technical) guideline is that the LLMs can "answer" anything you just gave them the answer for. So you give it a problem, ask it how to solve it, tell it to use that method and explain the data it needs, give it that data, and then show it everything at once: "With this data you requested and summarized, use this technique to answer this question".

Are you on other social media BTW? Thx.

Afraid not, only here. You can at me at [email protected] though.

> I've seen it invent ideas that (most-likely) were not in his training set.

How can you possibly know what's in its training set? Its training set is so large that any "guess" about what is or is not in it is a fool's errand. Given that we know how the technology works, there's no reason to claim emergent behavior when Occam's razor would say the items are likely in the training set.

It's not a fools errand, you can't know definitively but you can have a level of confidence.

It's quite easy to invent a sentence that's never been uttered before, it's also fairly easy to hit the limits of the internet's knowledge in pretty much any complex discipline that's not software engineering.

And it's apparent when you speak to ChatGPT, he will hallucinate a lot to compensate for niche topics it doesn't know much about.

Occam's razor wouldn't point to ChatGPT having omniscient knowledge of all things ever thought of and that will ever be invented.

This kind of stuff clearly shows that GPT can generalize concepts:

https://twitter.com/emollick/status/1626084142239649792

And this kind of stuff shows that it clearly cannot:

https://acoup.blog/2023/02/17/collections-on-chatgpt/

> If you know nothing about either book, this answer almost sounds useful (it isn’t). Now this is a trivial research task; simply typing ‘the limits of empire review’ into Google and then clicking on the very first non-paywalled result (this review of the book by David Potter from 1990) and reading the first paragraph makes almost immediately clear the correct answer is that Isaac’s book is an intentional and explicit rebuttal of Luttwak’s book, or as Potter puts it, “Ben Isaac’s The Limits of Empire offers a new and formidable challenge to Luttwack.”

> A human being who understands the words and what they mean could immediately answer the question, but ChatGPT which doesn’t, cannot: it can only BS around the answer by describing both books and then lamely saying they “intersect in some ways.”

> The information ChatGPT needed was clearly in its training materials (or it wouldn’t have a description of either book to make a lossy copy of), but it lacks the capacity to understand that information as information (rather than as a statistically correlated sequence of words). Consequently it cannot draw the right conclusion and so talks around the question in a convincing, but erronous way.

That appears to show that it can't answer some obscure trivia question to the standards of the blog post author. Nothing to do with generalizing concepts.

It's not an obscure trivia question. It's a question that anyone with an understanding of the material in question should be able to answer, with something vastly better than a superficial 'the two books are similar, but, like, different and talk about different stuff'.

It immediately cuts to the heart of the matter - an LLM does not understand things. It does not know anything but which word follows another. It can't do anything but regurgitate things it has read in an incredibly lossy manner.

But it's a neat party trick that seems to fool a lot of people into thinking that there's something legitimately useful, here.

Large language models are not search engines, even though a lot of people seem determined to treat them as such.

I don't think that blog post's excercise is a particularly useful demonstration of anything.

> You're also not right that GPT can't produce anything novel. Here's a tiny, modest example

Actually it turned out it didn't invent any of the games in that article. You can just Google them and find out.[0][1]

Reading is not believing.

0: https://www.digitaltrends.com/gaming/sumplete-chatgpt-ai-gam...

1: https://www.novelgames.com/en/labyrinthsudoku/

> I won't waste my time, as I see you didn't read what you responded to the first time. But you need to start paying attention. You denial & empty confidence in what's happening will only get you so far.

Maybe he didn't read what you wrote, but you didn't even bother Googling the claims in the article you posted.

I wonder how ChatGTP reinvented its own game? That indicates it's being retrained very frequently on crawl data which I wasn't aware was happening - certainly not with a cycle time in the weeks!

It didn't. The article is nonsense. He probably primed it beforehand by talking about Sudoku, or it may have just chanced onto something similar. When I ask it to code Sumplete, it creates a "math-based game in which players have to solve a sequence of simple addition problems within a limited time to progress to the next level."

It says:

> “Can you code a game called Sumplete?” Despite protesting about its inability to make games earlier, ChatGPT immediately started spitting out fully formed HTML, Javascript, and CSS code. Sure enough, it had once again created a version of Tait’s game – only this time, we never discussed what it actually was beforehand.

I guess "a version of Tait's game" leaves some wriggle room about exactly what it did.

The article seems to leave for itself wiggle room in several such places, suggesting the author had an agenda. Likely ideological rejection of AI as intelligence.

There is no evidence at all that it copied the smartphone game. No one has even been able to show a textual description of its rules online that could be copied. And modifying the rules of Sudoko to create a variation is well within ChatGPT's known abilities. You can easily test this stuff yourself.

It's quite bizarre to watch this discrimination against AI forming, where we make these increasingly more intelligent entities, and now everyone is out to prove they're stupid, and by definition they'll always be stupid, because only humans can be smart.

I don't want or need to anthropomorphize AI (yet?), but this reaction by us is a tad too similar to rather uncomfortable parts of our history.

Did you read carefully?

Your second link is a game different than Sumplete's rules, so not sure why you linked it.

You have one source, the Digital Trends article. I don't put that much weight on Sumplete in particular, but because I can't quote and sum up my experience of it producing quite original responses to my specific queries sometimes (next to, of course, quite a lot of mundane, trivial ones), so I thought I'd link to something objective.

Still the DT article speaks about a mobile game being similar. The link here is at best circumstantial. The app page doesn't describe the rules of the game. Instead it says this:

"It is very similar to classical game of Kakuro"

I checked Kakuro, and it has similar, but distinctly DIFFERENT rules, as well. So, what's the theory here? GPT downloaded the game and ran it on its phone? No.

Maybe the description of this game is somewhere. Maybe. But I couldn't find it. Where is it?

The excuse "it's somewhere, so it didn't invent it" will always be used for why an AI can't produce a "novel idea". While the reality of novel ideas is very simple: it's like conservation of matter & energy. You can never make something or destroy it. You can only rearrange it. New ideas are a rearrangement of other existing ideas. It's the case for everything we ever made. And everything we ourselves are.

This is a discussion in the art world that you'd hear a lot, before stable diffusion came out. The whole idea of "steal like an artist".

It's one of those questions about the nature of us that's on the level of "is there a ghost in the machine"; "do we have actually novel ideas?".

I'm kind of leaning the way you are; I don't think we do. What seems to be happening that we _call_ novelty is just an advanced form of recursive synthesis - we take information in and remix it. But we're capable of levels of abstraction, so we can remix it really well.

I think that getting the abstraction part into these models might just be as simple as wiring them together with a meta-model. A network responsible for identifying similarity between outputs of different types, or one that serves as a connective layer, distributing tokens through other networks and then synthesizing the results.

What I'm almost certain of, is that the important changes are going to be architectural.

The important changes will be architectural, yes.

Three things I'm noticing:

1. We're reinventing structured programming in AI in fast forward. First it's a plain Markov chain. Then it's an "attention" directed acyclic graph. Then we realize we need loops. Then we realize we need to jump to different points in the loop. Then we realize it's useful to recursively call yourself or parts of yourself as a subroutine, parametrized with specific input. Etc.

2. Even before we fully realize this framework of thought into a model, I'm almost sure the model EVOLVES some of these structures during training. In the form of crude unrolled loops etc. Simply because it's inevitable for processing certain types of input data.

3. In order to preserve pragmatic outcomes, I'd bet the future is not one giant monolithic model for AI, but many medium-sized models, communicating in a meta network, like meta neurons, sending meta (high-level) messages to each other.

Essentially, we need to make neural networks more like a fractal. I have this rule of thumb that always works somehow: "no concept definition is complete, until it's made recursive". Neural networks will get there.

For reference, I enjoyed playing sumplete. Master level passed the time nicely (not the full 20 hours, but I did come back to it!).

Mechanics work well for an AI generated game. No idea if it's novel or not, but it's a good take.

I think people also often figure out exactly what they think mid-sentence, by way of actually putting it into words. LLMs can iterate on their ideas as they generate the text.

I didn't agree with the OP but just noting this is wrong:

> These nets regurgitate the most average possible output given inputs. By definition.

You seem to be confusing the training objective with the capabilities.

For example:

> Why Can GPT Learn In-Context? Language Models Secretly Perform Gradient Descent as Meta-Optimizers

https://arxiv.org/abs/2212.10559

How do you get from bigger LLMs being better at few shot learning to being able to generate novelties?

I don't know.

But note that this is happening at inference time, so it's not just generating the average of the input the model is trained on.

Amen. Listen to this person, this is the voice of reason.

What do you think of this?

https://redsails.org/dialectics-and-ai/

How can you say the outputs are average when you don't know the weights.

Try getting a human to come up with a novel idea and you'll find their idea is just a remix of existing ones, partial understanding and a masked view of the world from their unique perspective (or they incorporated a random stimulus).

This constant reach to see human cognition as somehow unique (without evidence) is tiring and frankly an ego-extension / "made in God's image" type thought hole.

Do you think this net can tell us something about math or physics that we don’t know? If so, how?

The person making the claim that this is as or more intelligent than humans should at a minimum show it do something as impressive as humans. I’ll wait for a great piece of art or science until I for no reason diminish humans, that’s all. It’s not tiring to me to see the beauty of human intelligence.

People seem to be able to draw outside the lines. It’s cool, it’s beautiful. I shouldn’t have to point to any specific amazing achievements because there are too many. I haven’t seen one from an LLM yet. I’ll know it when I see it! Not hating, just clarifying.

I didn't say an LLM is as intelligent as humans.

I'm saying that human intelligence isn't some mystic process above replication or emulation ... and that if an LLM-type computation can do what it does now I see not reason it can't match or surpass human intelligence in the future.

It may produce that intelligence in a different way to our organic squishglobes but I don't see that it matters.

The vast majority of humans I've met have never produced great works of art or science. We're holding the models to higher standards than ourselves.

On this site lately there seems to be a truly deep sense of unmitigated woo over all things AI and this comment nicely demonstrates it.

We know so far that human cognition and sense of self is still largely a mystery and no evidence exists to show that it's just mechanical remixing of absorbed things. Contextually, many humans can indeed create novel suggestions, arguments and ideas, and they can self direct towards these in ways that no AI can so far. This does indeed make our cognition visibly unique without even having to mention anything religious

ChatGPT on the other hand is literally just a well-engineered design for creating coherent phrases from a huge training set of human information. As even its very creators admit, it doesn't consciously think, consider for real or perform literal AGI in any way that's sentient. For you and others here to call this equal to human cognition is absurd and not based on measurable evidence. It seems more guided by emotional awe at something presented in a new way than guided by sober reasoning. And you talk about human-centric ego arguments being tiring?

> We know so far that human cognition and sense of self is still largely a mystery and no evidence exists to show that it's just mechanical remixing of absorbed things

I'm not saying that ChatGPT is comparable in either capability or design to a human's cognition. What I am saying is that a human brain is just input,process and output. There's no reason a model of sufficient size can't emulate human thought in such a way that there's not much difference in capability (even if the mechanics differ).

I also think you're overstating the uniqueness of "novel" suggestions from humans and that "novel" ideas are some kind of high-water mark of intelligence rather than a fuzzy mechanism with induced feedback.

I can think of a novel idea by randomly mixing concepts I know. What matters is if this idea is good, and that test is done via interacting with the world. This mechanism can be implemented in an LLM now.

> You literally couldn’t get GPT to come up with a single novelty if you tried. It’s all remixing existing content, and again, doing so in a way to fit the average of the dataset.

So if I ask chatGPT to make up a poem about the power of friendship, and it must involve a unicorn and a jackelope, it will give me that. What's not novel about it? Isn't it pretty much what any person would do, mash up some related words?

What does novelty mean in this context?

Leaving aside the question of whether that combination of images is novel… In your example, all the proposed novelty is specified in the prompt. ChatGPT didn’t come up with it, you did.

> Isn't it pretty much what any person would do, mash up some related words?

Yes, that’s exactly what chatgpt does, and it’s what many humans do. But to be analogous to your initial example, there should be another person who actually came up with the instruction specifying which words to mash into a poem. The word-masher, whether human or chatgpt, is just following the instructions, not coming up with them.

The point is that it's able to produce content that hasn't been made before, it's not just giving back examples from the training set. It's able to--yes, with instruction--combine concepts logically to produce something new.

If I ask a random person I meet to write a poem with the same instructions as were given above I'd be surprised if they were able to come up with something as good as ChatGPT.

So kinda like when your English teacher asks you to write a poem about whatever you want? Can't ChatGPT do that as well? There's at least novelty in whatever random content it generates, is there not? Just by having a large enough library you will create a new thing because nobody's been in the jakelope/unicorn area before, or whatever the dice decide. I feel like maybe there's something about this no-novelty argument, but it's not been worded in a convincing way yet.

By this logic, all my farts are novel content.

I just gave chatGPT the prompt "write a poem" and it gave me an original little ditty about nature.

The novelty was not in the prompt.

> You literally couldn’t get GPT to come up with a single novelty if you tried

That's an interesting claim I don't know how to measure.

>> Hey, ChatGPT, could you invent five words that mean hungry?

> Sure, here are five words that mean hungry:

> 1. Cravacious

> 2. Rumble-hungry

> 3. Chewyemptiness

> 4. Famishy

> 5. Voracihunger

Is that novel? I don't know. It feels a lot like the word games Lewis Carroll or Roald Dahl played, though.

Few of those have zero results in Google search and sound fit for the purpose.

Unfortunately none of this matters, because "no true novel AI" is at this point just the "no true Scotsman" fallacy over and and over again.

We'll need to learn the hard way. Thanks for the creative test.

If you're looking for ways to measure creativity, ask it for recipes and then ask it to adapt them. It'll give you a recipe that uses strawberries and anchovies if you ask it. But like the words, it's remixing existing content and knowledge. Is that creative? I don't know. It's what a chef does, though.

> You literally couldn’t get GPT to come up with a single novelty if you tried.

Have you actually tried? You'd be surprised.

Here's a thought. Recently the whole connectome of the fruit fly has been mapped. 3000 neurons and 548k connections.

The fruit flys brain is capable of piloting the body of the fruit fly on a precision and energy efficiency that's simply unmatched in any kind of aerial machine we have developed. And only with those many connections, half a million? Neurons are very complex computation machines that can module their response very carefully.

The human brain does so much more than simply being able to process language. It's capable of one shot learning and to course correct incredibly easy. These features are very important for survivability and I would say are the "magic sauce" that we just haven't been able to replicate.

So I'm actually not so sure this is simply a parameters game. There could be important structures and specialisation that we are missing. In the fruit fly there are 93 different types of neurons, in the human brain iirc it might be hundreds.

A language model is also capable of one shot learning. Try it in the chat, you tell it something, once, and it incorporates it for the rest of the chat. :-)

You may think I don't know what "one shot learning" means, but I do know, I'm just saying it's a problem of our perspective, not of the model's capability. Aside from it having no long term memory (by design) from the chat. Yet.

Our high-level learning is absolutely nothing like how we learned as toddlers. It took us DECADES to understand complex concepts from "few shots". Only pretrained models can do that, and you're also one, in your adulthood.

What you say about extra complexity from neuron types and neurotransmitters combining with one another to modulate even simple networks is true of course. But it's also true you don't need to replicate all those organic details in a NN, because they're equivalent to just a few more "regular" parameters. Think of an artificial neuron in a network as a Turing machine. A Turing machine can do everything with just some more tape, and a neural network can do everything with just some more parameters.

And yet, our models show that we don't need AS many parameters as we thought. And that's exciting.

half a million connections is about 500 times more connections than there are in a 6502, so structure clearly matters a lot

> It has no baggage, it's built from the ground up to be a modern 21st century human. We're not, we train decades to become ones.

The whole point of machine learning is to do all that "decades" of training in a few hours.

In terms of learning material, it takes LLM a few orders of magnitude more material to learn than it does a human. After all, a human child trained with very few books (by age 7 a child would have seen only a few books) still outperforms CHatGPT trained with the same material, because ChatGPT trained on 4 books is basically useless.

They train on years of real-time streaming video. Feedback loops all over the place. Higher order structures and rules all get learned way, way before they even are able to touch a book. I cannot phantom the amount of data and analyses that passes a typical human child in a day.

So, sure, they can learn fast after all that. But so can GPT, which also one-shots a lot.

> They train on years of real-time streaming video.

...

> I cannot phantom the amount of data and analyses that passes a typical human child in a day.

You're counting every second of every image that a child sees as a separate "image" that is learned. I'm counting the number of objects that a child sees, because people aren't doing frame-capture and analysis on every image, they're doing lazy interpolation (which is why all your peripheral vision in in monochrome but you perceive it as full color).

A 6 month old baby who has only ever seen close family (parents, siblings, pets) is quite able to discern the same features on strangers.

Nothing we have in AI right now can take 3 example models of humans, and detect hair, nose and mouth in a brand new never before seen model of a human.

Not really the same thing as training on millions of different images and videos. The number of distinct and different objects in a 7 yo child's training set is rarely the same as "every single youtube video ever uploaded".

Going back to the original point: ChatGPT needs millions of books to learn to regurgitate information as well as a semi-intelligent young adult, while typical toddlers are quite able to regurgitate what they've just seen.

Humans and animals get by with millions of times less training data than AI models need.

>You're counting every second of every image that a child sees as a separate "image" that is learned. I'm counting the number of objects that a child sees, because people aren't doing frame-capture and analysis on every image, they're doing lazy interpolation (which is why all your peripheral vision in in monochrome but you perceive it as full color).

Then it sounds like he is slightly overestimating the dataset (since some of the frame is monochrome) and you are massively underestimating it.

Faces may not be the best example, because there's a lot of evidence babies are 'hardcoded' to be interested in and pay attention to faces. This implies there's some knowledge of them built in.

However, if you substitute abstract shapes (blocks, balls, etc.), your point absolutely stands up.

In some ways, though, isn't it more interesting to have an intelligence that learns in a completely different way from humans, as opposed to one that learns through essentially the same method?