AI 'breakthrough': neural net has human-like ability to generalize language

https://www.nature.com/articles/d41586-023-03272-3

by drcwpl • 3 years ago

204 71 3 years ago

drcwpl• 3 years ago

By contrast, GPT-4 struggled with the same task, failing, on average, between 42 and 86% of the time, depending on how the researchers presented the task. “It’s not magic, it’s practice,”

samfriedman• 3 years ago

Link to the actual paper: https://www.nature.com/articles/s41586-023-06668-3

mistrial9• 3 years ago

splendid sales headline - however, human communication is varied and multi-layered. Since there are too many angles to summarize -- suffice it to say that IMHO a healthy skepticism is warranted, along with evidence and test results.

link to cited repo: https://github.com/brendenlake/MLC/tree/v1.0.0

tomohelix• 3 years ago

I will take this article with a bag of salt but I think we are very close to solving the human communication problem. Not because the machines are perfect but because the human, our benchmarks, are prone to mistakes and so a "close enough" machine would pass as a human just fine.

marysnovirgin• 3 years ago

There is inherently no solution to the human communication problem. The mistake has always been in thinking technology would unite us and improve communication when clearly it's doing the opposite.

Beijinger• 3 years ago

I agree. But "Nature" is quite a name.

matt123456789• 3 years ago

And desperate to retain their relevance in the face of the onslaught of advancements that open access papers has unleashed.

valine4• 3 years ago

For anyone wondering about the architecture this is a 1.4 million parameter transformer model with both an encoder and decoder. The vocabulary size is comically small, it only understands 8 words.

It learns new ideas with very few examples, to the extent you can express new ideas with a vocabulary of eight words.

Izkata2• 3 years ago

Morse code has a vocabulary of 3 words (dot, dash, pause between human words).

yawnxyz• 3 years ago

those are more like tokens, not words

IMTDb• 3 years ago

Then English has a vocabulary of 27 words: A -> Z + space

rpearl2• 3 years ago

let's see...

English speakers tend to have a vocabulary of 30k words or so depending on what exactly you measure.

GPT-4 has 1.7 trillion parameters.

Of course scaling up is quite unpredictable in what capabilities it gets you, but It's not that much of a stretch to imagine that a GPT-4 sized model would have a reasonably sized vocabulary. Certainly worth testing if you have the resources to train such a thing!

glitchc• 3 years ago

It's the mapping of parameters to vocabulary claim embedded in your response that needs validation. 1.7 trillion parameters means what exactly?

Let's start with working vocabulary. Working vocabulary doesn't just mean knowing n words. It means putting n words together in factorially many valid combinations to construct sentences. And 30k btw is an insane working vocabulary. Most people know 1000 words on average in English. All of their sentences are structured from those 1000 words. This is true for most cases, except certain ones like Mandarin or German, where basic words can be used to assemble more complex words.

Certainly GPT-4 knows something. Presumably that something can be mapped to a working vocabulary. How large a vocabulary that is requires a testable, reproducible hypothesis supported by experimental proof. Do you have such a hypothesis with proof? Does anyone? Until we do it's just a guess.

swyx• 3 years ago

if you have the resources to train such a thing!

estimates i've heard for training gpt4 was $500m all in for what its worth

vessenes2• 3 years ago

I actually read the (first half) of the paper.

I can't speak to whether or not Nature should publish this, but I don't find the outcomes in the paper spectacular. In brief, they create an ontology -- in the initial examples, four made up color words, each corresponding to a real color and then three made up 'function' words, representing 'before', 'in the middle of' and 'triple'.

They then ask humans and their AI to generate sequences of colors based on short and progressively longer strings of color words and functions.

The AI is about as good at this as humans, which is to say 85% successful for instructions longer than three or four words.

The headline says that GPT4 is worse at this than MLC. I am doubtful about this claim. I feel a quality prompt engineer could beat 85% with GPT4.

The claims in the document are that MLC shows similar error types to humans, and that this backs some sort of theory of mind I don't know anything about. That's as may be.

I would be surprised if this becomes a radical new architecture based on what I'm reading. I couldn't find a lot of information as to size of the network used; I suppose if its very small that might be interesting. But this read to me very like an 'insider' paper, its target other academics who care about some of these theories of mind, not people who need to get sentences like 'cax kiki pup wif gep' turned into real color circles right away.

famouswaffles4• 3 years ago

I feel a quality prompt engineer could beat 85% with GPT4.

I generally don't pull the "you're not prompting it good enough" card unless i have direct experience with the task but this does raise a few flags - "between 42 and 86% of the time, depending on how the researchers presented the task."

I would have liked them to show how they presented this task, at least something more than a single throw away line, given how integral it is to the main claim.

randcraw• 3 years ago

The article suggested that GPF-4 FAILED between 42 and 86 percent of the time. So its success rate would be 58 to 14 percent, which compared to the novel NN's 80 percent, seems significant.

PumpkinSpice2• 3 years ago

I don't find this argument persuasive. If you need to engineer the prompt in order for the model to give you the answer you want, and if it's performing poorly without that kind of assistance from a human who understands the task and the desired outcome... then it's perfectly OK to not bother and give the LLM a low mark on a benchmark. After all, the whole point was to test the LLM's language and reasoning skills.

"Prompt engineering" makes sense if your goal is more utilitarian, but it's not a license to cheat on tests. If you want a machine learning model to generate aesthetically pleasing images, or to follow certain rules in a conversation, then it's OK to rely on hacky solutions to get there.

vessenes2• 3 years ago

Agreed. And there's going to be some significant domain decisions for GPT-4 to consider: Will the made up words be existing single tokens? Will those single tokens be heavily overloaded tokens, e.g. a letter of the alphabet? Will the representation of the circles be, say "B" for blue, or something else?

Along with this are questions as to whether you're going to treat GPT-4 as a zero shot or multi-shot oracle; while they have this idea of 'context free' challenges in the paper, they, crucially, train their network first.

Anyway, I like this paper less the more I think about it.

pama2• 3 years ago

The supplement discusses how they presented the task. Notably, they first gave all the training examples and then told the model what the task is and didn't ask it to reason or create any context before spitting out the answers. So basically the simplest way to interact with it, but probably not a great way to get solutions for this problem if that was the task at hand.

https://static-content.springer.com/esm/art%3A10.1038%2Fs415...

avgcorrection• 3 years ago

That’s seriously it? What an insulting (to the reader) headline by Nature.

8bitsrule• 3 years ago

I couldn't find a lot of information as to size of the network used

In the original paper,[0] in the section called 'Implementation of MLC', there's a description of sorts (Greek to me):

"Both the encoder and decoder have 3 layers, 8 attention heads per layer, input and hidden embeddings of size 128, and a feedforward hidden size of 512. Following GPT63, GELU64 activation functions are used instead of ReLU. In total, the architecture has about 1.4 million parameters."

[0] https://www.nature.com/articles/s41586-023-06668-3

sashank_1509• 3 years ago

Honestly I’m surprised by how low Nature has fallen. Is it even a useful signal of paper quality anymore? NeurIPS and ICLR aren’t great but in general I’ve found their work to be more rigorous than that of Nature despite of the fact that they are shorter conference papers compared to Nature’s journal papers.

Right now sadly the only useful signal in Deep Learning research is research group. If OpenAI releases a paper I know it’s something good that works at scale, similarly if Kaiming He, Piotr Dollar and team in Meta AI release a paper, it tends to be really good and SOTA. Google Deepmind maintains high quality, Google brain has been more of a mixed bag. If Berkeley releases a paper, it has 50% chance of directly going to trash, Stanford has a much lower percent (I also disambiguate) based on specific groups. Of course I’m going to be heavily biased and this system is not great, but conferences and journals have managed to become such a useless signal that I find this method to be more accurate.

tokai• 3 years ago

The claims in the document are that MLC shows similar error types to humans, and that this backs some sort of theory of mind

The same thing was claimed with latent semantic analysis in the late 90s.

jug• 3 years ago

But isn’t that their point? That their new neural network needs less prompt crafting because it is part better at reading intent and meaning of words and part less likely to hallucinate?

Sure, I also think GPT-4 can be brute forced with a very precise prompt into generating correct answers but that’s not how you would describe the problem to a human, and I assume the humans performing well in this test weren’t either, but given a fairly informal description of the exercise.

SirMaster• 3 years ago

If I am evaluating how intelligent an AI is against a human then IMO it's only fair they both get the same prompt.

It should be counted as a fault of the AI if it can't understand the prompt as well as the human, or another AI. That's all part of being a useful entity, how well you understand a prompt and how well you can infer the original meaning from what prompt you have been given.

MattPalmer10862• 3 years ago

I am no expert in this field, but this seems an important result, even if on a limited model.

I don't think that anyone was very sure if generalisation was actually happening with current architectures. Happy to be corrected by anyone who knows more.

MattPalmer10862• 3 years ago

Instead of downvoting, maybe someone can explain why this is not important?

m3kw9• 3 years ago

When an important word is quoted, don’t take it seriously

danielmarkbruce4• 3 years ago

This seems dumb. Sure, an LLM can't learn a new word. An LLM isn't an entire system. You could make a system (which extensively used an LLM) to do this fairly easily.

valine4• 3 years ago

LLMs can absolutely learn new words. Try it with GPT4, you can make up words with definitions and it will start using them in conversation.

danielmarkbruce4• 3 years ago

Learn implies change, but it's stateless.

What you are seeing a nice trick from the chatGPT application where they are keeping the history of your conversation (not in it's entirety, but a sort of summarized version), and in a single turn feeding it all back in.

In a single turn it can be told to use word x to mean y and generate output, but I don't think that's "learning".

matmulbro• 3 years ago

check out other articles by the same author: https://www.nature.com/search?author=Max+Kozlov

caddemon2• 3 years ago

This guy is a science journalist, the authors of the source research paper are more relevant if you're trying to infer anything about the material from the authors: https://www.nature.com/articles/s41586-023-06668-3

FWIW I think even the original paper is more of a master class in riding an academic bubble than anything of important substance. I'm just not sure what looking at other articles by this journalist is supposed to convey.

CSMastermind• 3 years ago

The link is to a news article about a recent paper not the a link to the research itself.

The editorialization of this being a breakthrough was done by the reporter not the authors of the paper.

I see people here comparing this to GPT-4 but I can't find that section in the actual paper nor in the underlying data. Can someone point me in the right direction?

pama2• 3 years ago

The fourth sentence of the discussion points to the relevant work in their supplement. Here is the link to the supplement: https://static-content.springer.com/esm/art%3A10.1038%2Fs415...

jmsuth• 3 years ago

sounds lit

kunley• 3 years ago

The question is, what kind of neural network was that. I mean, for a start, one of https://en.m.wikipedia.org/wiki/Types_of_artificial_neural_n...

The article only says "To make the neural net human-like, the authors trained it to reproduce the patterns of errors they observed in humans’ test results.", this is not much.

I'd especially like to know the deciding factor why it worked while LLM failed.

deadly_syn• 3 years ago

I have a theory we can get the same levels of output with a optimized pipeline of various specialozed chain. Output is forced into a ruleset to mimick english grammar. Hpt even seems to say this pattern has a lot of potential

AI 'breakthrough': neural net has human-like ability to generalize language

Related Stories