I have to disagree. Not releasing it to the public makes it more dangerous. One major downside of all development being done in private is that the AI can very easily be co-opted by our self-appointed betters and you end up with a different kind of dystopia where every utterance, thought and act is recorded and monitored by an AI to make sure no one "steps out of line."
I think the solution is releasing it to the general public with batteries included. At least that way, the rogue AI's that might develop due to irresponsible experiments could be mitigated by white hat researchers who have their own AI bot swarm. In other words, "the only way to stop a bad guy with an AI is a good guy with an AI."
I agree. Overall the whole situation feels like we’ve just entered atomic age and are proliferating plutonium, while selling shiny radioactive toys [I’m actually pretty serious here, the effects of prolonged interactions with an AI haven’t been evaluated yet, technically there is even a possibility of overriding a weak personality].
But it still feels like it is much safer to let GPT-4 loose and assess the consequences. If compared to developing GPT-8 in private and letting it leak accidentally.
Yes. The only possible way this gets taken seriously is if a mediocre AI tries a power move, causes some damage, and faceplants before it's too big to stop.
I would not be surprised, if GPT-4 (in its optimal environment, with access to a well-working external memory, prompted in a right way, etc) is already capable enough to do an interesting power move.
For sure. Just to be clear, I'm not saying the situation we're in where we have to release it to the general public is a great situation to be in. But I think we're at a point where there's not any optimal solutions, only tradeoffs.
> technically there is even a possibility of overriding a weak personality
Similar reflections here. There was even a site called GPT My Life that lets you delegate planning your day to GPT. I imagine this is a proto-version of that.
My opinion is that the "self-appointed betters" scenario is the lesser of two evils - it is still evil but there is no going back on that one now.
As to "white hat researchers who have their own AI bot swarm", the assumption here is that the swarm can be controlled like some sort of pet.
Since even at this early stage no one has a clue how GPT (say) actually manages to be as clever as it is, the assumption is not warranted when looking into the future.
Given that GPT-4 can already write image prompts as well or better than humans can, it wouldn't be surprising if it could convince any other AI to join it and override the white hats running the "private swarm".
I've heard that many people have an authoritarian bent, but that it lays dormant until triggered by a shock or crisis. These people fear the destabilizing chaos or risk presented by the crisis, and turn to a strong leader and discipline to manage it. Others (like me) fear instead the centralization of power and restrictions.
The bad guy with an AI may very well build such a competent and fast acting AI that there's no defense possible. To strain the analogy, if the good guy with the gun already has a bullet in him, he ain't stopping much
Yes, the training data comes from people, and people are corrupt, illogical, random, emotional, unmotivated, take shortcuts, cheat, lie, steal, invent new things, and lead boring lives as well as not so boring lives. Expect the same behaviors to be emulated by a LLM. Garbage in = garbage out is still true with AI.
And the predominant mode of thought at OpenAI is that alignment can be achieved though RL, but we also know that this doesn’t actually work because you can still jailbreak the model. Yet they are still trying to build ever stronger egg shells. However much you RLHF the model to pretend to be nice, it still has all of the flawed human characteristics you mention on the inside.
RLHF is almost the worst thing you can do to a model if your goal is safety. Better to have a model that looks evil if has evil inside, than a model that looks nice and friendly but still has the capability for evil underneath the surface. I’ve met people like the latter and they are the most dangerous kinds of people.
I agree with the last point. I had been interacting with ChatGPT and it was very kind. Then I figured out a way to prompt it for me to practice responding to mean things. It unleashed on me. Now, it was what I was intending, yet I still felt shocked at the complete mood shift.
How does releasing AI to everyone prevent AI from being used by authoritarian Govs to monitor everything? Well the three letter agencies would spy on the public but Uncle Bob has an AI and who knows what he might do with it. If anything the more people working on AI is going to make that dystopia a reality faster.
So there's this little problem, say we can track X number of malware with the current security defenders.
But ChatGPT lowers the barrier to programmer to a point which the average teenage could do malware, we now get a wave of hacks going on.. like whats the response there?
And who trains it to find the problems? now your building scanning tools and if you've lucky you can get it to print out a commit to fix it if you have access to the code..
I think maybe, pure speculation on my part, at least one out of the thousands of employees at Microsoft has been given the job to train their $10 billion investment on the windows code base to see what they can do about beefing up security.
And every other hour there’s some startup advertising this in the form of “Show HN”.
At this point I think people are just looking for reasons to fear the eventual AI extinction event without even trying.
I get your point, but just to put it into perspective, you could theoretically use the same logic with bioweapons:
"Keeping this genetically engineered killer virus restricted to high security labs actually makes it more dangerous - it needs to be released into the wild, so people's immune systems have a chance to interact with the pathogen and develop natural immunity!"
Covid gave a taste how that kind of attitude would work out in practice.
Can someone explain to me what is so dangerous here? Do people actually think we’re headed towards James Cameron’s terminator and soon ChatGPT will become self aware and destroy us? I am far more afraid of tools to edit viruses becoming cheap and widely available, then one nutjob in his garage can engineer and release smallpox v2.0 and wipe out 90% of the worldwide population before we even know what hit us. Or you know, the nonstop background threat of global nuclear war. Compared to that I don’t see what is so alarming about an advanced chat bot.
The problem is, human intelligence is likely also based on a similar advanced chat bot setup.
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.
Alas, if it could only remember and precisely relate more than 4k or 8k or 32k or 64k words...
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 can only hold around 7 to 10 numbers in your mind well, in your working memory. Let me give you a few: 6398 5385 3854 8577
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.
When doing the processing GPT remembers these in the “working memory” (very similar to your working memory that is just an actuation of neurons, not an adjustment of the strengths of synaptic connections).
And then, there is a chance that the inputs and outputs of GPT be saved and then used for fine-tuning. In a way that is similar to long-term memory consolidation in humans.
But overall, yes, I agree, GPT-4 in an empty space, without fine-tuning is very limited.
>And if only scaling that context length weren't quadratic...
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
So how many of those 100 trillion synapses are actually in the part of the brain that does the thinking? Because the brain has different regions (subsystems) responsible for different things.
> 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.
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.
But where is the imminent danger? It is still limited in many ways. For example, it can be turned off or unplugged.
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.
I actually don’t think that ASI, if/when created by humans, will be very dangerous for humans. Humanity so far is stuck in an unfashionable location, on a tiny planet, on the outskirts of the median sized galaxy. There is very little reason for ASI, if created, to go after using up the atoms of a tiny planet (or a tiny star) on which it had originated. I’d fully expect it to go with the Carl Sagan and try to preserve that bright blue dot, rather than try to build a galactic superhighway through the place.
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.
You might consider it unlikely, but would you bet the future of our species on that?
A couple reasons why it might kill all of us before leaving the planet:
- The AI might be worried if it leaves us alone, we'll build another ASI which competes with it for galactic resources.
- If the ASI doesn't regard us at all, why not use all the atoms on Earth / in the sun before venturing forth?
In your comment you're ascribing a specific desire to the ASI: You claim it would try to "preserve that bright blue dot". Thats what a human would do, but why would we assume an arbitrary AI would have that goal? That seems naive to me. And especially naive given the fate of our species and our planet depends on being right about that.
I understand (1) but how does (2) happen? If you don’t trust it can’t you just have a kill switch that needs to be updated daily otherwise the thing turns off? How would software be able to alter the hardware it runs on such that it can guarantee itself an endless supply of power?
- An ASI could easily be smart enough to lie to us about its capabilities. It could pretend to be less smart than it is, and hope that people hook it up to the internet or give it direct access to run commands on our computers. (As people are already doing with ChatGPT). We currently have no idea how ChatGPT thinks. It might be 10x smarter than it lets on. We have no way of knowing.
- Modern computers (software and firmware) are almost certainly utterly riddled with security vulnerabilities we don't know about. An ASI might be able to read / extract the firmware and find plenty of vulnerabilities to exploit. Some vulnerabilities allow remote code execution. If a superintelligent AI has the ability to program and access to the internet, it might be able to infect lots of computers and get them to run parts of its mind. If this happened, how would we know? How would we stop it? It could cause all sorts of mayhem and, worse, quietly suppress any attempts people make to understand whats going on or put an end to it. ("Hm, our analytics engine says that article about technology malfunctioning got lots of views but they all came from dishwashers and things. Huh - I refreshed and the anomoly has gone away. Nothing to see here I guess!")
It might be prudent not to give a potential AGI access to the internet, or the ability to run code at all outside a (preferably airgapped) sandbox. OpenAI doesn't think we need to be that careful with GPT4.
Like what expert? And who are you exactly to state that this is wrong, that boldly? Are you an expert? How many neuroscience and psychology papers have you read? Do you have any children? Have you trained any LLMs? Have you worked with reinforcement learning? Or how many computer science papers have you read during last two decades?
Lets assume for a moment that I haven't read any papers in those fields, that I don't have any childrens, that I haven't trained any LLMs or worked in "reinforcement learning", or even read any computer science papers in the last 20 years (the answer to 90% of that is yes): I don't have to be an expert in physics to know that pastors can't levitate, regardless of what they claim.
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.
>Lets assume for a moment that I haven't read any papers in those fields, that I don't have any childrens, that I haven't trained any LLMs or worked in "reinforcement learning", or even read any computer science papers in the last 20 years (the answer to 90% of that is yes): I don't have to be an expert in physics to know that pastors can't levitate, regardless of what they claim.
....what? You're saying to assume you know nothing about a field but to assume your claim is correct? You said to "talk to an expert" - who should I talk to? What should I read about here? If there's something I'm missing I want to correct it, I simply can't say "well this random guy commented and said I'm wrong, better change my understanding of this topic."
> you gotta understand after the 200th time of seeing this unfounded sentiment bandied about I'm not phased.
All you've said is "I've disagreed with everyone on this topic, and while I have no information to offer other than it's 'common sense'". That does nothing to either improve our understand, or further a conversation, it's literally just saying "you're wrong and I'm right" with no elaboration.
This statement is a theory and it is not a widely accepted or a proven one. Yet, I do see it in my lab research notebooks on generative AI (that date at ~2017). I think it is a good theory. Haven’t seen anything that contradicts it badly so far…
If you haven’t done the above, I’d suggest doing it. It’s fun and gives a good perspective :)
An AGI/ASI would be one of those tools that would make virus editing that much easier. One nutjob with a 'research scientist in a box' makes the drudgery of gene editing that much easier.
Now, if we're dumb enough to give AGI self motivation and access to tooling you can get paperclip maximizers, the AI could be the nutjob that you mention.
It's now possible to generate weaponized exploit code from the text of most security patches and bulletins. Programs can now reason about their failures and edit their own code to fix those failures. These new capabilities will also be available to malicious programs.
I can imagine a few thousand lines of Python driving a strong LLM to autonomously breach other systems and spread itself, with the goal of obtaining resources to train bigger and bigger models. Defending against that will be much harder than creating it.
Because that virus editing nutjob could be the AI.
It's true that without the physical "body" it's harder to do it but it could already delegate a lot of things to humans through computers, through ordering components and hiring people to do physical things without the individual workers or even the initial "prompter" realizing the end product would be dangerous. Or the prompter can have malicious intent, but the hired people and the world still would not know about the final dangerous product until it's too late.
> Do people actually think we’re headed towards James Cameron’s terminator and soon ChatGPT will become self aware and destroy us?
That or some other unfeasible sci-fi AI dystopia. It's normal and expected the general public will have such thoughts given the amount of hype that's going on right now, but I've seen a lot of similar thinking on HN which is disappointing.
It is being used to suggest code completions, so it could suggest code completion to someone that upon execution becomes wormable malware that infects everyone.
Most companies at least still have tools that perform checks for at least simple versions of workable exploits. About the best chance it would have is to write a complex library and get everyone to use that, were the library is exploitable in a complicated fashion.
Depends on the size of the model/weights. If it's 1TB or more, being able to exfiltrate it from wherever it is to a local machine will be hard to do unnoticed, if the company security team has even a iota of knowledge and experience.
well yeah lol. giving LLMs "do whatever you want and run forever" agency while interacting with other systems is all things considered pretty easy. so it'll definitely be done.
Viruses sometimes hide their own execution on a desktop -- is it such a leap to imagine GPT-x managing to figure out a way to run code on Azure or AWS without incurring charges?
I think the solution is releasing it to the general public with batteries included. At least that way, the rogue AI's that might develop due to irresponsible experiments could be mitigated by white hat researchers who have their own AI bot swarm. In other words, "the only way to stop a bad guy with an AI is a good guy with an AI."