OpenAI's statement on "governance of superintelligence" is basically "in our humble opinion you should regulate AI, but not *our* company's AI which is good, but instead only imaginary evil AI that exists only in the nightmares you have after reading too many Sci-Fi books, and the regulatory framework you choose should be this laughably guaranteed-to-fail regulatory framework which we designed here on a napkin while laughing" https://openai.com/blog/governance-of-superintelligence
Governance of superintelligence

Now is a good time to start thinking about the governance of superintelligence—future AI systems dramatically more capable than even AGI.

Like ... having been involved in major governments trying to set up coordinated action through multilateral organizations, my careful and considered response to any tech bro who says "major governments around the world could just set up [X]" is "lol, lmao"
I was going to say it's about the same amount of work trying to upgrade your biplane into a jumbo jet while you're flying it, but then it occurred to me that's actually considerably *more* tractable.
@Pwnallthethings Sounds like they could really use the help of a superintelligent AI to oversee all the different government organizations.
@jvert @Pwnallthethings And keep track of all the paperclips.
@habermas @jvert @Pwnallthethings "AI, we're out of paper clips. Get us more."

@Pwnallthethings Like repairing the Titanic, with the engines still running full steam, because drydock? What's a drydock?

Or at least, that's how I describe trying to reform the American healthcare system, which might be a slight outlier...

@Pwnallthethings i saw a talk at a con once where a guy who was ex IC said the majority of any op they do is by a very large margin, mostly documentation and bureaucracy. like taking weeks to get authorization to book a flight
@Viss Yeah, the baseline amount of work required to even be able to even start doing the basic job stuff in the job title is totally nuts.
@Pwnallthethings i barely know anything myself, but when i see regular people screaming shit about people tailing them and planting bugs in their houses, i just chuckle because my first thought is always "if they had ANY IDEA how much that would cost, and how many people it would take"
@Viss it takes a village hahaha
@Viss @Pwnallthethings these same ppl, they have mobile phones? Alexa? Ring? They use the Internet?
@Viss @Pwnallthethings this person has to be US ... it's not as bad elsewhere. But your point stands for sure
@Pwnallthethings “we could collectively agree” 😍🥰😘😂🤣😭
@Pwnallthethings yeah this post seems pretty out of touch, which is a bit more disappointing seeing as they are the market leader in this domain.
@Pwnallthethings I think tech bros overestimate the power of governments the same reason they overestimate the power of DAOs. They’re not good at understanding or relating to people, so it’s easier to think about them using really simple models. Works for computers, not so much for people.
@Pwnallthethings It took nearly 20 years to negotiate a UN treaty of the high seas. And that’s for a problem that’s literally bounded and nobody is giving up existential military or economic advantage.

@Pwnallthethings @briankrebs I don’t know, we could point to the United Nations and the decades of peace it ushered in.

Hahahahahahahahahaha

@Pwnallthethings Saved the best for last:
"Second, we believe it would be unintuitively risky and difficult to stop the creation of superintelligence. Because the upsides are so tremendous, the cost to build it decreases each year, the number of actors building it is rapidly increasing, and it’s inherently part of the technological path we are on, stopping it would require something like a global surveillance regime, and even that isn’t guaranteed to work. So we have to get it right." 🤡💨

@misc @Pwnallthethings They are demanding a public policy solution to a problem they claim is inevitable? I...

Look, I can understand speedrunning policy problems, I get it, but intentionally creating them by deliberately introducing market failures before the market has fully formed?

And these are the so-called tech libertarians...I am mildly amused.

@misc @Pwnallthethings i think you mean, a global surveillance regime that Silicon Valley doesn't run
@Pwnallthethings let's see one of these companies release a comprehensive human rights impact assessment on their *current* products. until then, laughable is indeed the right way to describe this - except that members of congress are slobbering all over these ppl.
@Pwnallthethings I keep thinking about the thing that said AI companies are like people selling radium cures saying we need to be very careful about people developing superpowers and meanwhile everyone's dying of cancer.
@Pwnallthethings gawd, what a bunch of wankers! 🤣

@Pwnallthethings Thank you, I needed a laugh. The first paragraph had major sophomore dorm rediscovering Malthus vibes. It either did not disappoint, or went down hill from there, depending on psrspective.

I was laughing enough I had to stop after the first few sections out of fear I'd drop my phone.

How do I put this in techbro? "Dude, do you even policy?" Is that how the youths say it these days?

@Pwnallthethings
@SwiftOnSecurity
I thought their advocating for regulation was just a way of making it expensive for newcomers, and pull the ladder up behind them.
@FritzAdalis @Pwnallthethings @SwiftOnSecurity This, and also making it so governments throw up their hands and say "YOU DO IT" and they say "Gladly..."
@Pwnallthethings And this is scary as hell, to have any corporation even advising governments instead of scientists or even individual workers on AI.
@Pwnallthethings “don’t require us to get permission to use our training data, because the robots are coming”

@Pwnallthethings “We think it’s important to allow companies and open-source projects to develop models below a significant capability threshold, without the kind of regulation we describe here (including burdensome mechanisms like licenses or audits).”

Licenses and audits, oh! The burden!

@Pwnallthethings so dumb, hs. We got the reality where the 4 horsemen suffer from the donner krueger effect.

@Pwnallthethings it's like we're sleep walking into Person of Interest - a show which one seemed at once possibly prescient but likely fantastical.

And yet, here we go.

@Pwnallthethings frankly gpt3.5 could have done a much better job.

@Pwnallthethings "we believe [AI is] going to lead to a much better world than what we can imagine today (we are already seeing early examples of this in areas like education, creative work, and personal productivity)"

We're seeing what now? I know of no benefits in education, creative work is worse off because artists are being pushed out by flawed okay-if-you-don't-look-too-hard creations, and "personal productivity" probably means "replacing people at work/expecting infinitely more" 😐

@Pwnallthethings
I can't see any way to do this except by criminalizing use cases, which in turn is complicated by the fact that most of the outrage is from people who see their jobs threatened, the more outraged the more suitable their job is for an AI.

And whatever regulations liberal democracies settle on, China, Iran, russia, North Korea will just ignore, so if the regulation limits competitiveness it helps authoritarian states.

@Pwnallthethings "Pay no attention to the techbros behind the curtain"
@Pwnallthethings "written on the back of a cigarettte packet"