> try to pressure the government into making it illegal for you to compete with us.
I mean the guy who created GPT-4 literally demanded a ban of any system more powerful than GPT-4.
I mean the guy who created GPT-4 literally demanded a ban of any system more powerful than GPT-4.
What I've seen from the horse's mouth is more like:
"""There are several other areas I mentioned in my written testimony where I believe that companies like ours can partner with governments, including ensuring that the most powerful AI models adhere to a set of safety requirements, facilitating processes to develop and update safety measures, and examining opportunities for global coordination."""
Now, sure, I could probably take a malevolent moustache twirling villain monologue and use chatGPT to turn it into a bland and prosaic statement like that, or even something utterly milquetoast, but if I try to imagine Altman playing 5D chess with the aid of his own AGI, there's the much easier solution of… just not telling anyone you have GPT-4 in the first place while using its power to manipulate etc.
"First, it is vital that AI companies–especially those working on the most powerful models–adhere to an appropriate set of safety requirements, including internal and external testing prior to release and publication of evaluation results. To ensure this, the U.S. government should consider a combination of licensing or registration requirements for development and release of AI models above a crucial threshold of capabilities, alongside incentives for full compliance with these requirement"
https://www.judiciary.senate.gov/imo/media/doc/2023-05-16%20...
You feel that quotation justifies the adjective "complex"?
While I don't doubt that ITAR is especially complex, following legislation is pretty common as a basic requirement in software development — say GDPR, or tax rules, or COPPA, etc. — it really doesn't seem to me that this is pushing all that hard for anything specific and detailed enough to even justify the claim that they're calling for a specific level of complexity of legislation.
Ah, fair enough then; it seemed to me like you were saying it was excessive and bad.
The same dynamic that incentivized a mass rollout of these unaligned systems will be perfectly sufficient to incentivize mass rollout of stronger, also unaligned systems.
I'm not going to take his fears as gospel, as he's spent so long focussing on his fear there's a danger of availability heuristic/attention biases having him take the worst case for everything.
I'm still going to promote caution, as the more potent a tech the greater the downside of being wrong, but it currently looks like we can cooperate with each other in this IRL iterated prisoner's dilemma.
OpenAI needs to make sure they have a business model where they can charge enterprise fees and provide value to corporations, not directly to people. ChatGPT was merely a publicity stunt. I do believe they got scared of how much value people could derive independently from it tho, hence the "regulate this" pressure.
Why would OpenAI care about that?
No, it means safety, as in not giving out dangerous answers that get people killed.
They don't — 49%, and they get to buy themselves out of even that, and if they do that by eating MS's lunch? Then it would suck to be an MS shareholder.
And they've been cautious since before the investment, with the cautious announcement of GPT-2 a month before the change in corporate structure to allow investors.
But even if OpenAI were owned completely by MS, why should MS care? Disrupting all employment sounds like the raison d'être for most if not all of silicon valley.
At the very least, imagine all the employees MS could let go if GPT-4 was even 80% as good as the advertisements in the form of "buy my guide for how to use GPT to make a fortune with best selling books/award winning advertisements/fantastic business proposals" keep claiming.
> Why would you think rich American engineers are in general any more worried about the overall safety of the world then their own self interest?
One is a subset of the other.
People regularly disagree about how risky things are, but they don't generally pick "destroy all things" over "destroy that which is close to me" when they actually believe that's the real choice and not, say, a panicked Chicken Little (which is how I think LeCun sees Yudkowsky).
They are clearly trying to angle for maximizing the profitability of that establishment.
They're not acting like they're part of the American corporate establishment, and I don't understand why you think they are.
Rather than quote my other comment comparing against 3M, I'll link to it: https://news.ycombinator.com/item?id=36677182
Your average OpenAI machine learning expert cares plenty about not killing people, just like you do.
Literally taken, that is quite close to impossible.
It was news days ago of somebody who committed suicide after having an interaction with a bot about nuclear risks or similar.
To avoid that, the bot would have to be a high-ranking professional psychologist with an explicit purpose not to trigger destructive reactions.
And that would fail the nature of a "consultant", which is something "under best effort to speak the truth" - incompatible direction with "something reassuring".
Sounds fairly dooable, TBH, given what else they're doing reasonably well at.
> And that would fail the nature of a "consultant", which is something "under best effort to speak the truth" - incompatible direction with "something reassuring".
I don't think they're as incompatible as you appear to.
> reasonably well
But I wrote «high-ranking professional». «Reasonably well» does not cover the cases relevant to full safety.
And, by the way, such attitude will backfire heavily on false positives.
Anyway: the case presented is there to point to a few difficulties involved in the idea of «not giving out dangerous answers».