Having panicked regulators with science fiction, I hope he's not surprised when they take action.
If Sam's Gambit succeeded, OpenAI could have potentially been granted a near absolute monopoly with the legislative reach of the US government working to imperil competitors through a Gordian knot of rules and regulations which OpenAI could have been the primary creator of, perhaps even as the head of some sort private-public 'Artificial Intelligence Accountability and Trust Division.' It really just gives one that happy feeling of bureaucrazy mixed with dystopia.
[1] - https://www.wired.com/story/openai-ceo-sam-altman-the-age-of...
I see the HN headlines like "new model approaches GPT4 in benchmarks!" but they turn out to suck when I try them. What can I use today that comes close to GPT4? Certainly not Bard.
I generally run 2 tests to gauge:
1. Write me a Python script that blah blah blah
2. Write me the lyrics to a song about blah blah blah
Only GPT4 has given me impressive results. For #2 the other models manage to rhyme (except most local models), but the rhymes on other models are really basic compared to GPT4's.
OpenAI is durable only if GPT is middlingly differentiated. If it's uncompetitive, they lose. If it's vital, they face regulation and state-sponsored competition. (I personally advocate for a Heavy Press Program [1] for training American and allied AIs.)
I don't think this was ever realistic. (I don't think Altman is that politically naive.) Instead, it was marketing. It's saying you've got something so powerful it's dangerous and should be treated carefully. While everyone's debating how to treat it carefully, they concede in public that it's powerful.
Maybe he really believes that he's the only person alive who can safely usher in a new AI godhead, and he feels this should be self-evident to everybody once he explains it to them.
If you consider the leaked Google Engineer memo about open source models being a threat, regulation and LLM registration would be the strategy to construct a moat.
https://twitter.com/soumithchintala/status/16712671501017210... https://archive.li/rfFlW
I'm not sure the most canonical paper on mixture of experts but here's one possible:
Not op, but this is where a cheeky google got me.
Innovation! :)
It’s more likely that we will see superintelligence in our lifetime. And if the rate of progress does not slow it will be sooner rather than later.
My current estimate is parity by 2030 and superintelligence by 2035. Evidence from specialist AIs, e.g. Go, indicates that super AIs tend to occur soon after parity is reached. E.g. AlphaGo (parity) March 2016; Master (super) December 2016.
Not if we are on the wrong path, going in the wrong direction, which I think was at least partly the point of the comment to which you are responding.
There is a school of thought that something fundamental is being missed by modern AI and the (amazing) success of GPT3+ ironically risks directing us further down that wrong path at an accelerating pace.
But a collection of models, with some kind of vector databases(or something more efficient than that), being orchestrated by one or multiple master models.
A large LLM that knows pretty much everything there is to know about the world, seems like a good building block for an AGI, no matter how the rest is built.
LLM's could be like the Broca's and Wernicke's area of an AGI brain. working in unison with dozens of other parts.
> That’s possible but doesn’t look likely at this point.
Why doesn’t it seem likely?
We’ve been trying to do great things with AI for decades. We don’t seem to have an excellent grasp on why certain things work well or if the strategies we’re using can ultimately yield much better intelligence.
My impression is that we really don’t have a lot of control over progress and we could very likely hit walls and stall for many years without meaningful progress. What am I missing?
1. We are wildly underestimating the computation requirements.
2. There are theoretical roadblocks coming up such that even a very large number of smart people being paid to solve the problem won’t find a key sequence of ideas. Think Riemann Hypothesis, or Fermat’s Last Theorem, etc.
The counter-argument to (1) is that available computational resources are very high given the billions of dollars available. The one system we know to possess human-parity intelligence (the human brain) uses 12 watts and is not exactly a data center.
The counter-argument to (2) is that we’ve made faster than expected progress since the discovery of transformers, and we seem to be quite close already given the capabilities of GPT-4. Of course you don’t know that you’ve hit a roadblock until you hit it, but so far it’s been smooth sailing.
It even seems unlikely that eve after having invented/improved the LLM, that the other components are anywhere close to being developed.
If his statement were honest and not strategic, then he's utterly irrational and quite possibly self-destructive (even genocidal). One simply wouldn't continue in that line of research if one believes there was the risk of an omnicidal AGI being spawned. It's like something out of a bad Lovecraft story... if you think the spell will summon Cthulhu, the best thing to do is just stop.
>WHY YOU SHOULD FEAR MACHINE INTELLIGENCE
>Development of superhuman machine intelligence (SMI) [1] is probably the greatest threat to the continued existence of humanity. There are other threats that I think are more certain to happen (for example, an engineered virus with a long incubation period and a high mortality rate) but are unlikely to destroy every human in the universe in the way that SMI could. Also, most of these other big threats are already widely feared.
- Sam Altman, February 25, 2015
[1] https://blog.samaltman.com/machine-intelligence-part-1
[2] https://blog.samaltman.com/machine-intelligence-part-2
[3] https://www.businessinsider.com/openai-ceo-sam-altman-says-h...
> We also have a bad habit of changing the definition of machine intelligence when a program gets really good to claim that the problem wasn’t really that hard in the first place
We've done this so much recently that I'm now seeing rewritten definitions of "real" intelligence that most humans do not meet.
Chess is the obvious example. A machine capable of playing chess at a human level was supposed to indicate the advent of genuine artificial intelligence at one time. It wasn't that playing chess well means one is intelligent, but rather it was assumed it'd entail abstract planning, strategic thought, intuition, and creativity. Of course now we have software which can crush even a world champion, but none of those adjacent capabilities emerged at all.
And so I think it's also increasingly obvious that this is the same thing with chatbots. Many of us thought those 'surrounding capabilities' were finally here, even more so with OpenAI regularly demonstrating exceptional competence on a wide array of distinct metrics, such as performance on the Bar exam. But once you use the system for a while it becomes clear that its knowledge base is absolutely and unbelievably immense, but its 'understanding' of that knowledge is literally zero. It will arbitrarily create e.g. API calls that do not exist, mix up utterly simple concepts, and fail to learn from its mistakes in any meaningful way whatsoever.
I'm sure if you've used ChatGPT for anything you've run into the utterly annoying scenario of:
- "How do I [x]?"
- "Sure! That's easy, just do [A]."
- "No, you're hallucinating."
- "Oh sorry, thanks. You're right you actually need to do [B]!"
- "No, you're still hallucinating."
- "Oh sorry, you're right. You just need to do [A].
If a human, even a stupid human, acted in this way - you'd assume they were trolling you, especially one gifted with the ability for infinite perfect and complete recall.
*: (such as when GPT-4 describes what would be output by Python code that I write and feed it, if it was run, despite GPT-4 not having any access to a Python interpreter and therefore having to simulate what one would do with my code, and despite my code not being in its training set.)
> especially one gifted with the ability for infinite perfect and complete recall.
You know that LLMs don't have this, right? There is no database they have access to containing their training data. They just have the weights that were optimized in response to seeing that training data.
------
"I'm working with a new computer language. What would be the output of:
IsTrue is not true
If IsTrue is true then print IsTrue
If IsTrue is not true then print IsNotTrue"
------
It hemmed and hawed, and accurately describe what the program would do, which is basically just repeating back the last two lines to me. But refused to tell me what the output would be. When I demanded it tell me what the value of "IsTrue" would be, so I could figure out the output, I got:
"I apologize for the confusion, but as an AI language model, I don't have access to the specific values of variables in your code or the ability to execute code directly. In the given context, the value of IsTrue is not specified, so I cannot determine its exact value. It could be either true or a value that is not true, depending on how the variable is defined or assigned in your code."
I then gave it the exact same program in C#, and it unsurprisingly gave me not only a far more meaningful description of what the code does, but also the exact output - immediately.
Because OpenAI says there isn't, and there are open source LLMs which show the same (but less profound) ability.
> expert system effectively akin to an interpreter working behind the scenes?
You're describing reasoning again. No-one hardcoded an "expert system" for Python into ChatGPT. If it has become one, it is through reading about Python (and same for C#).
Also, expert system [1] needs not be in quotes. It's an 'AI' technology dating back to the 60s. It's essentially just a fancy term for hard-coding queryable domain specific knowledge into a system. As an aside, where has OpenAI claimed any of this is false? Or, for that matter, which open source LLMs produce code that's not completely buggered?
Love this sentiment, stealing it =)
Altman peddles crypto. Keep that in mind when giving him the benefit of doubt on integrity and profit motivation.
I did an internship with ElementAI a while back. They had a full lobbying division to "come up with a legislative framework with the government for ethical AI". This is quite literally just a business strategy where you get to write the rules because lawmakers see you as "their guys" while challengers are seen as reckless.
It was already a thing in 2019 and it definitely is still a thing. Sam Altman just made the gambit that congress would see him as the honest and prudent guy. Seems like that just didn't work.
There is (IMHO) no good reason to let OpenAI argue simultaneously for "this technology is too dangerous to be replicated" and for "... but if you have a credit card then you can use it right now".
But he's not. As Scott Galloway put it, he's raising a gun to grandma's head and screaming "stop me before I shoot her!" He's saying ethically-responsible things. But simultaneously acting in contrast to his words.
Altman is doing an excellent job as a CEO. But that doesn't make him an angel. Just look at Worldcoin, his crypto project, and tell me you see the marks of a humanitarian.
However, they shouldn't be overstated. OpenAI has had the obviously best LLMs for over 4 years now. They aren't a flash in the pan.
If he believed it was truly dangerous, I don't think he'd be working for a company that is advertising they are actually pursuing super-intelligence ?
Not saying you're wrong, but there is definitely some very noticeable dissonance in his messaging and in his actions.
You're right that there's some dissonance there, however. Especially since three of the leading labs (OpenAI, Anthropic, and DeepMind) were _all_ founded on the premise that they had to be the ones to get superintelligence right because the risks were so high.
I think he wants to scare people a little in a somewhat controlled way to onramp us to this new reality as fast as possible.
I have a belief, as a ML researcher, that there are two types of x-risk AI/ML researchers. True believers and hype-men. People do hype the x-risk because it is catchy and gets more people talking. The whole "no news is bad news" strategy, which even worked in recent elections. I think there's two dangers to these people: 1) they eventually turn into true believers (you say something enough, you start to believe it), and 2) it distracts from the risks of the current dangers. For Sam, I think he is a true believer and I'd expect this out of any CEO who spends significant amounts of time hyping people up to gather capital by promising a future AGI.
As for myself, I'm not buying the x-risk arguments. There's a lot I have to say and a lot of nuance, but to put it briefly I'll reference something Mitchel said. She noted how it seems rather unlikely that a super-intelligence, who outperforms us in every single way is also unable to understand intention behind instructions (meaning it doesn't understand us, so a contradiction in super-intelligence). The danger is handing things over to a model without human supervision and letting it hallucinate. So the danger is thinking it is smarter than it actually is, and then trusting it. Concentrating on ineffable abilities of super-intelligences distracts us from this danger, which already exists.
Because it's a very serious possibility(that the singularity could end humanity), and a significant part of the people who are serious about AI are extremely worried about alignment, for good reasons.
> I hope he's not surprised when they take action.
Surprised? Do you mean relieved?
Because 1) the government isn't going to do anything and 2) investors will think OMG this must be profitable $$$
Because he sold his soul to the Devil (M$) and is now pursuing regulatory capture in order to build a "Open"AI-controlled moat where no competitors can survive.
Because he doesn't believe it's Science Fiction, you are projecting.
You can't poke a stick at one head and think the others won't bite you.
Altman effectively said to every level of government "OpenAI is extremely dangerous, you'd better look out for us".
All the wisdom of Bilbo Baggins reaching into the trolls pocket.