While GPT-4 only performs as good as top-10th percentile of human students taking an exam (a professional in the field can do much more than this), it is notable that as a generalist GPT-4 would outperform such professionals. And GPT-4 is much faster than a human. And we have not yet evaluated GPT-4 working in its optimal setting (access to optimal external tools). And we have not yet seen GPT-5 or 6 or 8.
So, get ready for an interesting ride.
And if only scaling that context length weren't quadratic...
Indeed, we would really expect an AI to be able to achieve AGI. And it might decide to do all kinds of alien things. The sky would not be the limit!
We have more than 100 trillion synapses in our brains. That's not our "parameter" count. It's the size of the thing that's getting squared at every "step". LLMs are amazing, but the next valley of disillusionment is going to begin when that quadratic scaling cost begins to rear its head and we are left in breathless anticipation of something better.
I am not as worried, I guess, as your average AI ethicist. I can hope for the best (I welcome the singularity as much as the next nerd), but quadratic isn't going to get easier without some very new kinds of computers. For those to scale to AGI on this planet it's questionable if they'll have the same architecture we're working with now. Otherwise, I'd expect a being whose brain is a rock with lightning in it to have take over the world long, long ago. Earth has plenty of both for something smart and energy efficient to have evolved in all these billions of years. But it didn't and maybe that's a lesson.
That all said, these LLMs are really amazing at language. Just don't ask them to link a narrative arc into some subtle detail that appeared twice in the last three hundred pages of text. For a human it ain't a problem. But these systems need to grow a ton of new helper functionality and subsystems to hope to achieve that kind of performance. And, I'll venture that kind of thing is a lower bound on the abilitites of any being who would be able to savage the world with it's intellect. It will have to be able to link up so, so many disparate threads to do it. It boggles our minds, which are only squaring a measly 100T dimension every tick. Ahem.
You have 1 second, close your eyes and add them together. Write down the result.
I’m pretty sure that GPT-4 at its 4k setting would outperform you.
[The point being, we have not seen what even GPT-4 can do in its optimal environment. Humans use paper, computers, google, etc. to organize their thoughts and work efficiently. They don’t just sit in empty space and then put everything into the working memory and magically produce the results. So imagine now that you do have a similar level of tooling and sophistication around GPT-4, like there is present around humans. I’m considering that and it is difficult to extrapolate what even GPT-4 can do, in its optimal environment.]
I'll point out that chatgpt needs to be paying attention to the numbers to remember them in the way I'm taking about. You will need to fine tune it or something to get it to remember them blind. I suppose that's not what you're talking about?
There is a strong chance that I'll remember where to find these numbers in a decade, after seeing and hearing untold trillions of "tokens" of input. The topic (Auto-GPT, which is revolutionary), my arguments about biological complexity (I'll continue to refine them but the rendition here was particularly fun to write) or any of these things will key me back to look up the precise details (here: these high entropy numbers). Attention is perhaps all you need... But in the world it's not quite arranged the same way as in machines. They're going to need some serious augmentation and extension to have these capabilities over the scales than we find trivial.
edit: you expanded your comment. Yes. We are augmented. Just dealing with all those augmented features requires precisely the long range correlation tracking I'm taking about. I don't doubt these systems will become ever more powerful, and will be adapted into a wider environment until their capabilities become truly human like. I am suggesting that the long range correlation issue is key. It's precisely what uniques humans from other beings on this planet. We have crazy endurance and our brains both cause and support that capability. All those connections are what let's us chase down large game, farm a piece of land for decades, write encyclopedias, and build complex cultures and relationships with hundreds and thousands of others. I'll be happy to be wrong, but it looks hard-as-in-quadratic to get this kind of general intelligence out of machines. Which scales badly.
There are transformers approximations that are not quadratic (available out of the box since more than a year) :
Two schools of thoughts here :
- People that approximate the neighbor search with something like "Reformer" and O(L log(L) ) time and memory complexity.
- People that use a low-rank approximation of the attention product with something like "Linformer" with O(L) complexity but with more sensibility to transformer rank collapse
Agreed: LLM are just one of many necessary modules. But amazing nonetheless. The quadratic scaling problem needs an attentional-conceptual extractor layer with working memory. Hofstadter points out that this needs to be structured as a recursive “strange loop” (p 709 of GEB). Thalamo-cortico-thalamic circuitry is a strange loop and attentional self-control may happens by phase- or time-shifting activity of different circuits to achieve flexible “binding” for attention and compute.
I’m actually optimistic that this is not a heavy computational lift but a clever deep extension of recursive self-modulating algorithms across modules. The recursion is key. And the embodiment is also probably crucial to bootstrap self-consciousness. Watching infants bootstrap is an inspiration.
Is it because CAPTCHAs won’t work anymore? That sounds like a problem for sites like Twitter that have bot problems.
Is it because it may replace people’s jobs? That comes with every technological step forward and there’s always alarmist ludditism to accompany it.
Is it because bad people will use it to do bad things? Again, that comes with every new technology and that’s a law enforcement problem.
I don’t really see what the imminent danger is, just sounds like the first few big players trying to create a regulatory moat and lock out potential new upstarts. Or they’re just distracting regulators from something else, like maybe antitrust enforcement.
1. GPT-8 or something is able to do 70% of people’s jobs. It can write software, drive cars, design industrial processes, build robots and manufacture anything we can imagine. This is a great thing in the long term, but in the short term society is designed where you need to work in order to have food to eat. I expect a period of rioting, poverty, and general instability.
All we need for this to be the case is a human level AI.
2. But we won’t stop improving AIs when they operate at human level. An ASI (artificial superintelligence) would be deeply unpredictable to us. Trying to figure out what an ASI will do is like a dog trying to understand a human. If we make an ASI that’s not properly aligned with human interests, there’s a good chance it will kill everyone. And unfortunately, we might only get one chance to properly align it before it escapes the lab and starts modifying its own code.
Smart people disagree on how likely these scenarios are. I think (1) is likely within my lifetime. And I think it’s very unlikely we stop improving AIs when they’re at human levels of intelligence. (GPT4 already exceeds human minds in the breadth of its long term memory and its speed.)
That’s why people are worried, and making nuclear weapon analogies in this thread.
It’s the intermediate steps that I’m more worried about. Like Ilya or Sam making a few mistakes, because of lack of sleep or some silly peer pressure.
This is so wildly wrong and yet confidently said in every techbro post about LLMs. I beg of you to talk to an expert.
You're mad that I'm calling you out, I get it, but you gotta understand after the 200th time of seeing this unfounded sentiment bandied about I'm not phased.
