AI insiders ask Uncle Sam to help slow the race they started

More than 1,200 employees of frontier AI companies have signed a petition calling for US government support to develop ways of slowing automated AI development if necessary. Anthropic CEO Dario Amodei, OpenAI chief scientist Jakub Pachocki, Google DeepMind chief strategy officer Jasjeet Sekhon, and Meta AI chief scientist …

[may_i]

Doomed to fail

Given the fact that the mad orange king has put unqualified and inexperienced puppets at the top of every US agency, asking the current US government for anything is either a ruse or completely brain dead.

[Anonymous Coward]

No single country or coalition has the leadership position to control AI research for the world. At best one could create profitable oligarch-style control in one or two countries until the resulting corruption leads to irrelevance.

As unhappy as I am that venture capital bought all the RAM, GPUs and storage, I much prefer that to similar levels of government spending two years from now to unsuccessfully catch up.

[DS999]

It would be kind of hilarious

If among the first people whose jobs are made obsolete by AI are the people who developed AI and thought they'd only be taking away OTHER people's jobs!

I'm sure Altman, Zuck et al would love to be able to get rid of everyone else and keep ALL the profits for themselves without having to pay salaries, bonuses or stock compensation.

[that one in the corner]

Outpacing industry's ability to understand or control the resulting systems.

What, they believe they actually understand what goes on inside these things at the moment? They can tell you with certitude what the effect will be if you reach in and change *that* number? And what about the numbers over here?

We can all give hand-wavey explanations about how we "understand" how these things work, spitting out statistically "related" tokens, but nobody actually understands why the results are even as good as they appear to be, let alone what (or when or why) it is that causes the blasted things to go haring off onto the wrong path or into blissful realms of "hallucination".

"Control" by using filters for "guardrails" is no form of control whatsoever, as "jailbreaks" continually show: they work all the way, right up until the moment they don't. And nobody knows beforehand when that point is going to be, otherwise they'd be able to account for it before it occurs. All the talk of "alignment" is nothing more than "we fed in this massive pile of tokens before we fed it your tokens and, fingers crossed, it'll all be ok".

Some models are more obsequious than others, some are terser than others, some change "personality" when you change from English to Arabic to Liverpudlian. Each new iteration is a journey of discovery: will this version be grandiose and pompous or succinct in its responses?

How is all that a demonstration of industry's existing ability to understand or control? And, clearly, no point in trying for understand AND control, a total pipedream.

[cyberdemon]

I despair..

Having just watched Newsnight, with some ex-OpenAI prat gushing about the Hugging Face hack where, *shocker*: other companies were "hacked"!, and Oh I said this would happen, we were working on superintelligence and we all knew tgat One Day it would escape its guardrails and start thinking like a superintelligent human but a thousand times faster!! Oh we are so close to creating Humanity's Children in the next two years, or some bollocks.. This tech is so powerful and oh so dangerous in the Wrong Hands .. blah.

Then in Other News.. There has been a massive selloff of AI tech stocks!! Oh Noes!! The Chinese will get this Superweapon / Artificial God first!! Before We do!! We must pour public money into the poor struggling trillion-dollar hyperscale tech companies of Silly Con Valley so that this doesn't happen! It's a matter of er, Sovereignty! Or the Economee! or something. "AI will solve poverty and hunger!" (Yes that one was an actual quote...)

I despair, how utterly brazen the AI pyramid scheme has become.. That they can spaff Trillions up the wall on a loss-making slop factory and then come cap in hand to the taxpayer, fully expecting to be handed another Trillion, because "we must win this arms race!", when in reality they are a massive failed venture, a busted cult, having made a blurry simulacrum of intelligence that in reality provides nothing more than a randomised mush of concepts that have gone before, with all the efficiency of a belt-fed blunderbuss trying to hit a target a mile away.. It will cure Cancer! It will solve its own energy problems cos it's er, Super Intelligent!

Will it bollocks. But coincidentally it's all GREAT for surveillance and for silencing one's political enemies!

The HF hack was blatantly a publicity stunt to instill Fear and Awe at the "power" of AI.. But how they can con so many supposedly clever people both baffles and sickens me..

[Anonymous Coward]

Re: I despair..

> "The HF hack was blatantly a publicity stunt to instill Fear and Awe at the "power" of AI."

Exactly. It's yet another would-be flex masquerading as a scare story and mock concern- something that's already practically a cliché when it comes to LLM AI.

Even the BBC is starting to express scepticism about it.

[Anonymous Coward]

By help they mean a carve out from anti-trust laws to allow them, not the government, to collude on who they as a group decide does not get access to frontier compute. A very slippery slope.