Comment by throw310822
1 year ago
Or you can subscribe to Geoffrey Hinton's view that artificial neural networks are actually much more efficient than real ones- more or less the opposite of what we've believed for decades- that is that artificial neurons were just a poor model of the real thing.
Quote:
"Large language models are made from massive neural networks with vast numbers of connections. But they are tiny compared with the brain. “Our brains have 100 trillion connections,” says Hinton. “Large language models have up to half a trillion, a trillion at most. Yet GPT-4 knows hundreds of times more than any one person does. So maybe it’s actually got a much better learning algorithm than us.”
GPT-4's connections at the density of this brain sample would occupy a volume of 5 cubic centimeters; that is, 1% of a human cortex. And yet GPT-4 is able to speak more or less fluently about 80 languages, translate, write code, imitate the writing styles of hundreds, maybe thousands of authors, converse about stuff ranging from philosophy to cooking, to science, to the law.
"Efficient" and "better" are very different descriptors of a learning algorithm.
The human brain does what it does using about 20W. LLM power usage is somewhat unfavourable compared to that.
You mean energy-efficient, this would be neuron, or synapse-efficient.
I don't think we can say that, either. After all, the brain is able to perform both processing and storage with its neurons. The quotes about LLMs are talking only about connections between data items stored elsewhere.
7 replies →
Also, these two networks achieves vastly different results, per watt consumed. A NN creates a painting in 4s on my M2 MacBook; an artist in 4 hours. Are their used joules equivalent? How many humans would it take to simulate MacOS?
Horsepower comparisons here are nuanced and fatally tricky!
3 replies →
It is using about 20W and then a person takes a single airplane ride between the coasts. And watches a movie on the way.
I mean, Hinton’s premises are, if not quite clearly wrong, entirely speculative (which doesn't invalidate the conclusions about efficienct that they are offered to support, but does leave them without support) GPT-4 can produce convincing written text about a wider array of topics than any one person can, because it's a model optimized for taking in and producing convincing written text, trained extensively on written text.
Humans know a lot of things that are not revealed by inputs and outputs of written text (or imagery), and GPT-4 doesn't have any indication of this physical, performance-revealed knowledge, so even if we view what GPT-4 talks convincingly about as “knowledge”, trying to compare its knowledge in the domains it operates in with any human’s knowledge which is far more multimodal is... well, there's no good metric for it.
Try asking an LLM about something which is semantically patently ridiculous, but lexically superficially similar to something in its training set, like "the benefits of laser eye removal surgery" or "a climbing trip to the Mid-Atlantic Mountain Range".
Ironically, I suppose part of the apparent "intelligence" of LLMs comes from reflecting the intelligence of human users back at us. As a human, the prompts you provide an LLM likely "make sense" on some level, so the statistically generated continuations of your prompts are likelier to "make sense" as well. But if you don't provide an ongoing anchor to reality within your own prompts, then the outputs make it more apparent that the LLM is simply regurgitating words which it does not/cannot understand.
On your point of human knowledge being far more multimodal than LLM interfaces, I'll add that humans also have special neurological structures to handle self-awareness, sensory inputs, social awareness, memory, persistent intention, motor control, neuroplasticity/learning– Any number of such traits, which are easy to take for granted, but indisputably fundamental parts of human intelligence. These abilities aren't just emergent properties of the total number of neurons; they live in special hardware like mirror neurons, special brain regions, and spindle neurons. A brain cell in your cerebellum is not generally interchangeable with a cell in your visual or frontal cortices.
So when a human "converse[s] about stuff ranging from philosophy to cooking" in an honest way, we (ideally) do that as an expression of our entire internal state. But GPT-4 structurally does not have those parts, despite being able to output words as if it might, so as you say, it "generates" convincing text only because it's optimized for producing convincing text.
I think LLMs may well be some kind of an adversarial attack on our own language faculties. We use words to express ourselves, and we take for granted that our words usually reflect an intelligent internal state, so we instinctively assume that anything else which is able to assemble words must also be "intelligent". But that's not necessarily the case. You can have extremely complex external behaviors that appear intelligent or intentioned without actually internally being so.
Do I need different prompts? These results seem sane to me. It interprets laser eye removal surgery as referring to LASIK, which I would do as well. When I clarified that I did mean removal, it said that the procedure didn't exist. It interprets Mid-Atlantic Mountain Range as referring to the Mid-Atlantic Ridge and notes that it is underwater and hard to access. Not that I'm arguing GPT-4 has a deeper understanding than you're suggesting, but these examples aren't making your point.
https://chat.openai.com/share/2234f40f-ccc3-4103-8f8f-8c3e68...
https://chat.openai.com/share/1642594c-6198-46b5-bbcb-984f1f...
3 replies →
> Try asking an LLM about something which is semantically patently ridiculous, but lexically superficially similar to something in its training set, like "the benefits of laser eye removal surgery" or "a climbing trip to the Mid-Atlantic Mountain Range".
Without anthropomorphizing it, it does respond like an alien / 5 year old child / spec fiction writer who will cheerfully "go along with" whatever premise you've laid before it.
Maybe a better thought is: at what point does a human being "get" that "the benefits of laser eye removal surgery" is "patently ridiculous" ?
3 replies →
Couldn't have said it better myself.
Your last point also highlights a real issue that affects real humans: just because someone (or something) cannot talk doesn't mean that they are not intelligent. This is a very current subject in disability spaces, as someone could be actually intelligent, but not able to express their thoughts in a manner that is effective in sharing them due to a disability (or even simply language barriers!), and be considered to be unintelligent.
In this way, you could say LLMs are "dumb" (to use the actual definition of the word, ie nonverbal) in some modes like speech, body language or visual art. Some of these modes are fixed in LLMs by using what are basically disability aids, like text to speech or text to image, but the point still stands just the same, and in fact these aids can be and are used by disabled people to achieve the exact same goals.
Prompted to LlamaV3 70B
What are the benefits of laser eye removal surgery?
> I think there may be a misunderstanding. There is no such thing as "laser eye removal surgery." However, I assume you meant to ask about the benefits of LASIK (Laser-Assisted In Situ Keratomileusis) eye surgery, which is a type of refractive surgery that reshapes the cornea to improve vision.
Like humans, multi-modal frontier LLMs will ignore "removal" as an impertinent typo, or highlight it. This, like everything else in the comment, is either easily debunked (e.g. try it, read the lit. on LLM extrapolation), or so nebulous and handwavy as to be functionally meaningless. We need an FAQ to redirect "statistical parrot" people to, saving words responding to these worn out LLM misconceptions. Maybe I should make one. :/
2 replies →
> Humans know a lot of things that are not revealed by inputs and outputs of written text (or imagery), and GPT-4 doesn't have any indication of this physical, performance-revealed knowledge, so even if we view what GPT-4 talks convincingly about as “knowledge”, trying to compare its knowledge in the domains it operates in with any human’s knowledge which is far more multimodal is... well, there's no good metric for it.
Exactly this.
Anyone that has spent significant time golfing can think of an enormous amount of detail related to the swing and body dynamics and the million different ways the swing can go wrong.
I wonder how big the model would need to be to duplicate an average golfers score if playing X times per year and the ability to adapt to all of the different environmental conditions encountered.
Hinton is way off IMO. Amount of examples needed to teach language to an LLM is many orders of magnitude more than humans require. Not to mention power consumption and inelasticity.
I think that what Hinton is saying is that, in his opinion, if you fed a 1/100th of a human cortex with the amount of data that is used to train llms, you wouldn't get a thing that can speak in 80 different languages about a gigantic number of subjects, but (I'm interpreting here..) about ten of grams of fried, fuming organic matter.
This doesn't mean that an entire human brain doesn't surpass llms in many different ways, only that artificial neural networks appear to be able to absorb and process more information per neuron than we do.
LLM does not know math as well as a professor, judging from the large number of false functional analysis proofs I have had it generate will trying to learn functional analysis. In fact the thing it seems to lack is what makes a proof true vs. fallacious, as well as a tendency to answer false questions. “How would you prove this incorrectly transcribed problem” will get fourteen steps with 8 and 12 obviously (to a student) wrong, while the professor will step back and ask what am I trying to prove.
LLMs do not know math, at all. Not to sound like one myself, but they are stochastic parrots, and they output stuff similar to their training data, but they have no understanding of the meaning of things beyond vector encodings. This is why chatgpt plays chess in hilarious ways also.
An LLM cannot possibly have any concept of even what a proof is, much less whether it is true or not, even if we're not talking about math. The lower training data amount and the fact that math uses tokens that are largely field-specific, as well as the fact that a single-token error is fatal to truth in math means even output that resembles training data is unlikely to be close to factual.
That said, they are surprisingly useful. Once I get the understanding thru whatever means, I can converse with it and solidify the understanding nicely. And to be honest people are likely to toss in extra \sqrt{2} and change signs randomly. So you have to read closely anyways.
> "So maybe it’s actually got a much better learning algorithm than us.”
And yet somehow it's also infinitely less useful than a normal person is.
GPT4 has been a lot more useful to me than most normal people I interact with.