AI security activist group PauseAI coordinated international protests to name for a pause in improvement of AI fashions extra highly effective than GPT-4.
Protestors got here collectively on Monday in 14 cities world wide, together with New York, London, Sydney, and Sao Paulo to lift consciousness of potential AI dangers.
The acknowledged objective of the protest in keeping with PauseAI’s web site is to “persuade the few highly effective people (ministers) who might be visiting the following AI Security Summit (the twenty second of Might) to be the adults within the room. It’s as much as us to make them perceive that the dangers are too massive to disregard, that they’re the one ones who’ve the facility to repair the issue.”
It’s occurring. Our second worldwide protest.https://t.co/FlF1y5jpO2 pic.twitter.com/RmOcRtbKiC
— PauseAI ⏸ (@PauseAI) Might 12, 2024
The AI Seoul Summit, which takes place on 21 and 22 Might, is the 6-month follow-up to the UK AI Security Summit held final November. The preliminary enthusiasm for worldwide cooperation on AI security appears to be wavering, with a number of contributors pulling out of the upcoming summit.
PauseAI is pragmatic in acknowledging that “we can’t anticipate nations or firms to danger their aggressive benefit by pausing AI coaching runs for a very long time if different nations or firms don’t do the identical. Because of this we want a worldwide Pause.”
They need the first objective of the upcoming summit to be the institution of a world AI security company, much like the IAEA. Satirically, that is one thing that OpenAI CEO Sam Altman additionally urged.
PauseAI’s protests calling for a halt to extra superior fashions occurred on the identical day that OpenAI launched GPT-4o, its quicker and extra highly effective iteration of its GPT-4 mannequin.
Whereas AI followers marveled over GPT-4o’s new capabilities, the PauseAI protestors have been much less enthusiastic over the exponential progress in new AI fashions.
Will sit-ins, placards, and handing out leaflets be sufficient to coax world leaders into taking the significant motion PauseAI says humanity wants? Can the inexorable march of AI developments be stopped lengthy sufficient to make sure that AI security guardrails are in place?
PauseAI thinks it’s potential. In a put up on X, the group mentioned, “AGI shouldn’t be inevitable. It requires hordes of engineers with million-dollar paychecks. It requires a completely purposeful and unrestricted provide chain of essentially the most complicated {hardware}. It requires all of us to permit these firms to gamble with our future.”
Protests have been efficient in bringing about motion in opposition to GMO meals, nuclear weapons, and local weather change. However, the inherent risks of these points are extra extensively accepted by world leaders.
The arguments over future AI questions of safety need to compete with rising pleasure over tangible AI advantages that humanity is experiencing proper now.
Will PauseAI protestors obtain their goal and earn humanity’s gratitude? Will their fears in the end show to be unfounded? Or will they be those that finally say “We advised you so.”
