Live data from Hacker News

Gavin Newsom vetoes SB 1047

gov.ca.gov

321–330 of 514 posts

Re: Gavin Newsom vetoes SB 1047

#321

Terrible piece of legislation. Glad the governor took it down. This is what regulatory capture looks like. Someone commoditized your product, so you make it illegal for them to continue making your stuff free. Might as well make Linux illegal so everyone is forced to use Microsoft and Apple.

I disagree.

On this topic, I’m seeing too many ideological and uninformed claims.

It is hard for many aspiring AI startup founders to rationally and neutrally assess the AI landscape, pros and cons.

Re: Gavin Newsom vetoes SB 1047

#322
post #214

Earlier quoted context omitted.

That’s why there are two thresholds.

Cost per FLOP continues to drop on an exponential trend (and what bit flops do we mean?). Leaving aside more effective training methodologies and how that muddies everything by allowing superior to GPT4 perf using less training flops, it also means one of the thresholds soon will not make sense. With the other threshold, it creates a disincentive for models like llama-405B+, in effect enshrining an even wider gap bet…

Why? Llama is not generated by some guy in a shed.

And even if it were, if said guy has such amount of compute, then it's time to use some of it to describe the model's safety profile.

If it makes sense for Meta to release models, it would have made sense even with the requirement. (After all the whole point of the proposed regulation is to get some better sense of those closed models.)

Re: Gavin Newsom vetoes SB 1047

#323

Earlier quoted context omitted.

> impossibly vague for a piece of legislation - "reasonable care" and "unreasonable risk" sound like things that could be endlessly litigated. Nope, that's entirely standard legal stuff. Tort law deals exactly with those kinds of things, for instance. Yes it can certainly wind up in litigation, but the entire point is that if there's a gray area, a company should make sure it's operating entirely within the OK area -…

> but "reasonable care" and "unreasonable risk" are not problematic. Still strongly disagree, at least when it comes to AI legislation. Yes, I fully realize that there are "reasonableness" standards in lots of places of US jurisprudence, but when it comes to AI, given how new the tech is and how, perhaps more than any other recent technology, it is largely a "black box", meaning we don't really know how it works and…

Exactly. It's about as meaningful as passing a law making it illegal to be a criminal. Right, so what does that actually mean apart from "we'll decide when it happens"?

Re: Gavin Newsom vetoes SB 1047

#324
post #320

The immediate danger of large AI models isn't that they'll turn the earth to paperclips it's that we'll create fraud as a service and have a society where nothing can be trusted. I'd be all for a law (however clumsy) that made image, audio or video content produced by models with over X parameters to be marked with metadata saying it's AI generated. Creating models that don't tag their output as such would be banned.…

I think the most acheivable way of having some verification of AI images is simply for the AI generators to store finger prints of every image they generate. That way if you ever want to know you can go back to Meta or whoever and say "Hey, here's this image, do you think it came from you". There's already technology for that sort of thing in the world (content ID from youtube, CSAM detection etc.). It's obviously no…

> That way if you ever want to know you can go back to Meta or whoever and say "Hey, here's this image, do you think it came from you".

Firstly, if you want to know an image isn’t generated, you’d have to go to every ‘whoever’ in the world, including companies that no longer exist.

Secondly, if you ask evil.com that question, you would have to trust them to answer honestly for both all images they generated and images they didn’t generate (claiming real pictures were generated by you can probably be career-ending for a politician)

This is worse than https://www.cs.utexas.edu/~EWD/ewd02xx/EWD249.PDF: “Program testing can be used to show the presence of bugs, but never to show their absence!”. You can neither show an image is real nor that it is fake.

Re: Gavin Newsom vetoes SB 1047

#326

Earlier quoted context omitted.

The future survival of humanity involves creating machines that have all of our knowledge and which can replicate themselves. We can't leave the planet but our robot children can. I just wish that I could see what they become.

Sure, that's future survival. Is it of humanity though? Kinda no by definition in your scenario. In general, depends at least if they share our values...

Values...values? Hopefully not, since they would be completely useless.

Re: Gavin Newsom vetoes SB 1047

#328
This veto document is shockingly lucid, I'm quite impressed with it despite my belief that regulation as a strategy for managing critical risks of AI is misguided.

tl;dr gavin newsom thinks that a signable bill needs "safety protocols, proactive guardrails, and severe consequences" based on some general framework guided by "empirical trajectory analysis", and also is mindful of the promise/threat/gravity of all the machine learning occurring in CA specifically. He also affirms a general appetite for CA to take on a leadership role wrt regulating AI. My general read is that he wants to preserve public attention on the need for AI regulation and not squander it on SB 1047 specifically. Or who knows I'm not a politician lol. Really strong document tho,

Interesting segment:

> By focusing only on the most expensive and large-scale models, SB 1047 establishes a regulatory framework that could give the public a false sense of security about controlling this fast-moving technology. Smaller, specialized models may emerge as equally or even more dangerous than the models targeted by SB 1047 - at the potential expense of curtailing the very innovation that fuels advancement in favor of the public good. > Adaptability is critical as we race to regulate a technology still in its infancy. This will require a delicate balance. While well-intentioned, SB 1047 does not take into account whether an Al system is deployed in high-risk environments, involves critical decision-making or the use of sensitive data. Instead, the bill applies stringent standards to even the most basic functions - so long as a large system deploys it. I do not believe this is the best approach to protecting the public from real threats posed by the technology.

This is an incisive critique of the fundamental initial goal of SB 1047. Based on the fact that the bill explicitly seeks to cover models whose training cost was >=$100m & expensive fine-tunes, my initial guess about this bill was that it was designed by someone software engineering-minded scared of e.g. open-weight releases a la facebook/mistral/etc teaching someone how to build a nuke or something. LLMs probably replaced the ubiquitous robot lady pictures you see in every AI-focused article as public enemy number one, and the bill feels focused on some of the technical specifics of this advent and its widespread use. This focus blinds the bill from addressing the general danger of machine learning however, which naturally confounds regulation for precisely the reason plain-spoken in the above four sentences. Incredible technical communication here.

Re: Gavin Newsom vetoes SB 1047

#329
post #320

The immediate danger of large AI models isn't that they'll turn the earth to paperclips it's that we'll create fraud as a service and have a society where nothing can be trusted. I'd be all for a law (however clumsy) that made image, audio or video content produced by models with over X parameters to be marked with metadata saying it's AI generated. Creating models that don't tag their output as such would be banned.…

I think the most acheivable way of having some verification of AI images is simply for the AI generators to store finger prints of every image they generate. That way if you ever want to know you can go back to Meta or whoever and say "Hey, here's this image, do you think it came from you". There's already technology for that sort of thing in the world (content ID from youtube, CSAM detection etc.). It's obviously no…

What's to stop someone from downloading an open source model, running it themselves, and either just not sharing the hashes, subtly corrupting the hash algo so that it gives a false negative, etc?

Also you need perceptual hashing (since one bitflip of the generated media alters the whole hash) which is squishy and not perfectly reliable to begin with.

Post reply on HN