Don’t believe the hype: AGI is far from inevitable
Don’t believe the hype: AGI is far from inevitable
Will AI soon surpass the human brain? If you ask employees at OpenAI, Google DeepMind and other large tech companies, it is inevitable. However, researchers at Radboud University and other institutes show new proof that those claims are overblown.
cross-posted from: https://lemmy.ml/post/20858435
Will AI soon surpass the human brain? If you ask employees at OpenAI, Google DeepMind and other large tech companies, it is inevitable. However, researchers at Radboud University and other institutes show new proof that those claims are overblown and unlikely to ever come to fruition. Their findings are published in Computational Brain & Behavior today.
Meh. It's not a problem of scale. It's a problem of we have no idea how the fuck to do that. Scaling up existing techniques is neither necessary nor sufficient.
39Right on the money. One of the big things with AI safety is "we have no fucking clue how AGI can originate so we are constantly in the dark." If we ever did create it, we likely would not immediately know it was AGI, and that creation could go very terribly in a number of ways.
5
Sounds really counterintuitive to say that it’s impossible.
The article says that we would run out of computing power, and that’s definitely true for current hardware and software. It’s just that they are being developed all the time, so I think we need to leave that door open. Who knows how efficient things can get within the next decade or century. The article didn’t even mention any fundamental obstacle that would make AGI completely impossible. It’s not like AGI would be violating the laws of physics.
24Whenever I hear someone say that something is impossible with current technology, I think about my grandma. When she was a kid, only some important people had telephones. Doctors, police, etc.
In her lifetime we went from that to today, and, since she's still alive, even further into the future.
Whenever someone calls something impossible, I think about how far technology will progress in my own lifetime and I know that they've got no idea what they're talking about. (Unless, like you said, it's against the laws of physics. But sometimes even then I'm not so sure, cause it's not like we understand those entirely. )
16The thing is, we have no idea where technological progress is taking us. So far, most predictions have been wrong. 50 to 60 years ago, people thought we would already be colonizing other planets by now. Barely anyone was able to predict the Internet, smartphones, social media, etc. - the kind of technology that is actually shaping our civilization's future right now.
Another aspect that I feel is often neglected is the assumption that technological progress will continue forever or at least continue at this current rapid pace. This wasn't true in the past and we might simply be experiencing a historical anomaly right now, one that could correct itself very soon in the future, either towards stagnation or even regression.
14Let's put it this way: If in our lifetime we can simulate the intelligence of a vinegar fly as general intelligence, that would be a monumental landmark in AGI. And we're far, far, far away from it.
As far as the iron age was from the metal alloys used in the Space Shuttle.
Talking about AGI simulating higher intelligence at the level of a dog or a cat, dear I say a pigeon or a crow is as far fetched as expecting ancient Egyptians to harness the power of the atom.
11Could take a while, but how long? Progress tends to be non-linear, so things can slow down and speed up suddenly. I’m pretty sure we’ll get there sooner or later unless we nuke ourselves to oblivion before that.
If AI development isn’t prioritized, it could take centuries. Maybe we’re still missing some crucial corner stores we haven’t even thought of yet. Just imagine what it was like to build an airplane in an age when the internal combustion engine hadn’t been invented yet. Maybe we’re still missing something that big. On the other hand, it could also be just around the corner, but I find it unlikely.
3
Actually, we do already know that we're close to a theoretical limit of increasing computing power as we currently know it. The transistor can't really get that much smaller, before it stops working.
Also, if you're talking about the article as linked, that is a mere introduction to a much longer paper.
6The fact that human brain is capable of general intelligence tells us everything we need to know about the processing power needed to run one.
5Well it sets an upper bound on compute requirements at 'simulate 10^27 atoms for thirty years' remains to be seen if what we can optimize away ever converges with what's feasible to build.
7
The article did mention a fundamental obstacle. It said quite clearly that we would run out of resources before we had enough computing power. I suppose you could counter that by arguing that we could discover magic, or magical technology, or a lot of new resources through space exploration.
Of course things get more efficient. But in the past few decades they've gotten efficient in predictable, and mostly predicted, ways. It's certainly possible that totally unexpected things can happen. I could win the lottery next week. Is that the standard? Are you pushing the stance that says AGI is somewhat less likely than winning the lottery or getting struck by lightning, but by golly it's more than zero, how dare you suggest that it's anywhere close to zero?
4It really depends on your assumptions. If you assume that software and hardware will stay at the current level, then the article does present a valid point. I would argue that those assumptions are only reasonable in the short term. AGI development does depend on some big technological changes we haven’t seen yet, so it could take decades or even a century, but I wouldn’t call it impossible.
If you assumed that 1950s style vacuum tube computers were the best thing ever, you could safely say that playing a game like fortnite with your buddies living in different countries is completely impossible. Modern semiconductors and integrated circuits would have seemed pretty magical in that context.
If we assume that we’re going to be stuck with silicon, you can safely say that AGI just isn’t going to happen with these tools and methods. Since quantum computers aren’t quite useful just yet and optical computers aren’t even in the news in any meaningful way, it seems that we will be stuck with silicon for quite some time. However, in the long term, you can’t really say that for sure. Technological developments have taken sudden and unpredictable jumps from time to time.
2
The actual paper is an interesting read. They present an actual computational proof, stating that even if you have essentially infinite memory, a computer that's a billion times faster than what we have now, perfect training data that you can sample without bias and you're only aiming for an AGI that performs slightly better than chance, it's still completely infeasible to do within the next few millenia. Ergo, it's definitely not "right around the corner". We're lightyears off still.
They prove this by proving that if you could train an AI in a tractable amount of time, you would have proven P=NP. And thus, training an AI is NP-hard. Given the minimum data that needs to be learned to be better than chance, this results in a ridiculously long training time well beyond the realm of what's even remotely feasible. And that's provided you don't even have to deal with all the constraints that exist in the real world.
We perhaps need some breakthrough in quantum computing in order to get closer. That is not to say that AI won't improve or anything, it'll get a bit better. But there is a computationally proven ceiling here, and breaking through that is exceptionally hard.
It also raises (imo) the question of whether or not we can truly consider humans to have general intelligence or not. Perhaps we're not as smart as we think we are either.
21The paper's scope is to prove that AI cannot feasibly be trained, using training data and learning algorithms, into something that approximates human cognition.
The limits of that finding are important here: it's not that creating an AGI is impossible, it's just that however it will be made, it will need to be made some other way, not by training alone.
Our squishy brains (or perhaps more accurately, our nervous systems contained within a biochemical organism influenced by a microbiome) arose out of evolutionary selection algorithms, so general intelligence is clearly possible.
So it may still be the case that AGI via computation alone is possible, and that creating such an AGI will not require solution of an NP-hard problem. But this paper closes one potential pathway that many believe is a viable pathway (if the paper's proof is actually correct, I definitely am not the person to make that evaluation). That doesn't mean they've proven there's no pathway at all.
10Our squishy brains (or perhaps more accurately, our nervous systems contained within a biochemical organism influenced by a microbiome) arose out of evolutionary selection algorithms, so general intelligence is clearly possible.
That's assuming that we are a general intelligence. I'm actually unsure if that's even true.
That doesn't mean they've proven there's no pathway at all.
True, they've only calculated it'd take perhaps millions of years. Which might be accurate, I'm not sure to what kind of computer global evolution over trillions of organisms over millions of years adds up to. And yes, perhaps some breakthrough happens, but it's still very unlikely and definitely not "right around the corner" as the AI-bros claim (and that near-future thing is what the paper set out to disprove).
4
A breakthrough in quantum computing wouldn't necessarily help. QC isn't faster than classical computing in the general case, it just happens to be for a few specific algorithms (e.g. factoring numbers). It's not impossible that a QC breakthrough might speed up training AI models (although to my knowledge we don't have any reason to believe that it would) and maybe that's what you're referring to, but there's a widespread misconception that Quantum computers are essentially non-deterministic turing machines that "evaluate all possible states at the same time" which isn't the case.
9I was more hinting at that through conventional computational means we're just not getting there, and that some completely hypothetical breakthrough somewhere is required. QC is the best guess I have for where it might be but it's still far-fetched.
But yes, you're absolutely right that QC in general isn't a magic bullet here.
8
Nitpick: a lightyear is a measure of distance, not of time :)
1Yes, hence we're not "right around the corner", it's a figure of speech that uses spatial distance to metaphorically show we're very far away from something.
4
AGI is inevitable unless:
-
General intelligence is substrate independent and what the brain does cannot be replicated in silica. However, since both are made of matter, and matter obeys the laws of physics, I see no reason to assume this.
-
We destroy ourselves before we reach AGI.
Other than that, we will keep incrementally improving our technology and it's only a matter of time untill we get there. May take 5 years, 50 or 500 but it seems pretty inevitable to me.
17@ContrarianTrail @JRepin well I guess somebody would first need to clearly define what "AGI" is. Currently it's just "whatever the techbro hypers want it to be".
And then there's the matter (ha!) of your assumption that we understand all laws of physics necessary that "matter obeys", or that we can reasonably understand them. That's a pretty strong assumption: individual human minds are pretty limited and communication adds overhead, and we might reach a point where we're stuck.
13A chess engine is intelligent in one thing: playing chess. That narrow intelligence doesn’t translate to any other skill, even if it's sometimes superhuman at that one task, like a calculator.
Humans, on the other hand, are generally intelligent. We can perform a variety of cognitive tasks that are unrelated to each other, with our only limitations being the physical ones of our "meat computer."
Artificial General Intelligence (AGI) is the artificial version of human cognitive capabilities, but without the brain's limitations. It should be noted that AGI is not synonymous with AI. AGI is a type of AI, but not all AI is generally intelligent. The next step from AGI would be Artificial Super Intelligence (ASI), which would not only be generally intelligent but also superhumanly so. This is what the "AI doomers" are concerned about.
11@ContrarianTrail @JRepin and finally, there's a question of whether we actually decide to pursue it.
Nuclear power was supposed to be the "inevitable" power source for all of humanity mere 50 years ago. But at some point we decided not to pursue that goal.
Cryptocurrencies were supposed to be "inevitable" replacement for the banking system.
And we *have* cryptocurrencies and nuclear power. These exist. As opposed to whatever nebulous concept hides beneath "AGI".
2
Another possibility is that humans just aren't smart enough to figure out AGI. While I'm sure that we will continue incrementally improving technology in some form, it's not at all self-evident that these improvements will eventually add up to AGI.
7I get what you're saying but to me, that still just sounds like a timescale issue. I can't think of a scenario where we've improved something so much that there's just absolutely nothing we could improve on further. With AI we only need to reach the point of making it have human-level cognitive capabilities and from there on it can improve itself.
1
-
Will AI soon surpass the human brain?
If you ask employees at OpenAI, Google DeepMind and other large tech companies, it is inevitable.That doesn't answer the question.
If it will happen is unrelated to When it will happen.
I'd expect we'll see AGI some time between the next 20 and 200 years. I think that's pretty soon. You may not.9If there were a giant asteroid hurling toward Earth, set to impact sometime in the next 20 to 200 years, I’d say there’s definitely a need for urgency. A true AGI is somewhat of an asteroidal impact in itself.
6None of those companies are suggesting 20 years. They're suggesting much less than 10, and selling investors on that promise.
2
Possible or not I don't think we'll get to the point of AGI. I'm pretty sure at some point someone will do something monumentally stupid with AI that will wipe out humanity.
6Like wrecking the biosphere in its persuit.
9Maybe. But I have a feeling it'll be a dumb single mistake that'll make someone say "ah, shit" just before we're wiped out.
When the Soviets trained anti-tank dogs in WW2 they did so on tanks that weren't running to save fuel: "Their deployment revealed some serious problems... In the field, the dogs refused to dive under moving tanks." https://en.m.wikipedia.org/wiki/Anti-tank_dog
History is littered with these kinds of mistakes. It would only take one military AI with access to autonomous weapons to have a similar issue in it's training data to potentially kill us all.
1
The steam engine won't replace John Henry!!!
6Not really a good comparison. The steam engine was an extant technology at that point. AGI is not, and we really no idea if/when it will be. One thing is clear though, it is not as close on the horizon as tech bros want us to think it is.
6
I like SCUMM but AGI is okay I just don't like typing commands
4