How the Loudest Voices in AI Went From ‘Regulate Us’ to ‘Unleash Us’

May 16, 2023, Sam Altman appeared Faced with a subcommittee of the judiciary of the Senate. The title of the hearing was “monitoring the AI”. The session was a lovefest, with Altman and the senators celebrating what Altman called the IA “moment of printing” – and recognizing that the United States needed solid laws to avoid its traps. “We believe that regulatory intervention by governments will be essential to mitigate the risk of increasingly powerful models,” he said. The legislators hung on to each word of Altman when it springs how intelligent laws could allow the AI ​​to flourish, but only in the directives of companies that the legislators and manufacturers of AI considered essential at that time. Altman spoke for industry, which largely shared his attitude. The battle cry “regulated us!”

Two years later, on May 8 of this year, Altman was back in front of another group of senators. The Senators and Altman still sang the same piece, but one fired a different playlist. This audience was called “to win the AI ​​race”. In DC, the word “surveillance” fell in disgrace, and the discourse of AI is no exception. Instead of pleading so that external organizations examine AI models to assess the risks, or for platforms to alert people when they interact with AI, the chairman of the Ted Cruz committee has pleaded for a path where the government would not only feed innovation but would remove barriers such as “overregulation”. Altman was on board with that. His message was no longer “regulate me” but “invest in me”. He said that excessive surregulation – like the rules adopted by the European Union or a recently bill Veto in California would be “disastrous”. “We need space to innovate and move quickly,” he said. The security railings could be necessary, he said, but they had to involve “a sensible regulation that does not slow us down”.

What happened? On the one hand, the time of panic just after everyone has panicked It has become clear This congress was not going to move quickly on the AI. But the greatest development is that Donald Trump took over the White House and hit the brakes in the nuanced and pro-regulator of the Biden administration. The doctrine of Trump of AI regulation seems to be concerned about that of Trump’s supporter, Marc Andreessen, who declared in his Optimistic techno manifest This AI regulation was literally a form of murder because “any deceleration of AI will cost lives”. Vice-president JD Vance made these priorities explicit in an international rally held in Paris in February. “I’m not here … to talk about AI security, which was the title of the conference a few years ago,” he said. “We believe that an excessive regulation of the AI ​​sector could kill a transformative industry just as it discolors, and we will do everything possible to encourage pro-growth policies.” The administration then unveiled a Ai Action Plan “Improve America’s position as the AI ​​power and prevent requirements unnecessarily contrary to the stack of private sector innovation.”

Two enemies emerged in this movement. The first is the European Union which adopted a regulatory This requires the transparency and responsibility of large AI companies. The White House despises this approach, just like those who build AI companies in the United States.

But the biggest couge is China. The perspective of the People’s Republic that defeated the United States in the “IA race” is so unthinkable that the regulations must be put aside, or done with what Altman and Cruz have described as a “light touch”. Part of this reasoning comes from a theory known as “Defense hard”, “ This postulates that the AI ​​models can reach a tilting point where self-abundance of fast lightning is launching a dizzying gyre of supercapability, also known as Ag. “If you get there first, your vile person, I will not be able to catch you,” said the former CEO of Google Eric SchmidtWith the “you” as a competitor (Schmidt had spoken of China’s status as a leader in open source.) Schmidt is one of the strongest votes warning of this possible future. But the White House is probably less interested in singularity than in conventional economic competition.

The fear that China is advancing on AI is the main engine of current American policy, security will be damaged. The party line is even opposed to individual states trying to fill the vacuum of inaction with their own laws. The version of tax creation, Megabill at Medicaid Cup has just passed by the house included a Mandated moratorium on any legislation on AI at the level of the State for 10 years. It is like eternity in terms of AI progress. (Experts say that this provision will not survive opposition to the Senate, but it should be noted that almost all the Republicans of the Chamber voted for this.)