However AI nerds might keep in mind that precisely a yr in the past, on July 21, 2023, Biden was posing with seven prime tech executives on the White Home. He’d simply negotiated a deal the place they agreed to eight of probably the most prescriptive guidelines focused on the AI sector at the moment. Rather a lot can change in a yr!
The voluntary commitments have been hailed as much-needed steerage for the AI sector, which was constructing highly effective know-how with few guardrails. Since then, eight extra firms have signed the commitments, and the White Home has issued an government order that expands upon them—for instance, with a requirement that builders share security take a look at outcomes for brand new AI fashions with the US authorities if the exams present that the know-how may pose a threat to nationwide safety.
US politics is extraordinarily polarized, and the nation is unlikely to go AI regulation anytime quickly. So these commitments, together with some current legal guidelines resembling antitrust and client safety guidelines, are the very best the US has when it comes to defending folks from AI harms. To mark the one-year anniversary of the voluntary commitments, I made a decision to take a look at what’s occurred since. I requested the unique seven firms that signed the voluntary commitments to share as a lot as they might on what they’ve carried out to adjust to them, cross-checked their responses with a handful of exterior consultants, and tried my greatest to offer a way of how a lot progress has been made. You possibly can learn my story right here.
Silicon Valley hates being regulated and argues that it hinders innovation. Proper now, the US is counting on the tech sector’s goodwill to guard its customers from hurt, however these firms can determine to alter their insurance policies anytime that fits them and face no actual penalties. And that’s the issue with nonbinding commitments: They’re simple to signal, and as simple to neglect.
That’s to not say they don’t have any worth. They are often helpful in creating norms round AI improvement and putting public stress on firms to do higher. In only one yr, tech firms have carried out some optimistic modifications, resembling AI red-teaming, watermarking, and funding in analysis on methods to make AI programs secure. Nevertheless, these types of commitments are opt-in solely, and which means firms can at all times simply choose again out once more. Which brings me to the subsequent huge query for this discipline: The place will Biden’s successor take US AI coverage?
The controversy round AI regulation is unlikely to go away if Donald Trump wins the presidential election in November, says Brandie Nonnecke, the director of the CITRIS Coverage Lab at UC Berkeley.
“Typically the events have completely different issues about using AI. One is likely to be extra involved about workforce results, and one other is likely to be extra involved about bias and discrimination,” says Nonnecke. “It’s clear that it’s a bipartisan situation that there should be some guardrails and oversight of AI improvement in the US,” she provides.
Trump isn’t any stranger to AI. Whereas in workplace, he signed an government order calling for extra funding in AI analysis and asking the federal authorities to make use of extra AI, coordinated by a brand new Nationwide AI Initiative Workplace. He additionally issued early steerage on accountable AI. If he returns to workplace, he’s reportedly planning to scratch Biden’s government order and put in place his personal AI government order that reduces AI regulation and units up a “Manhattan Venture” to spice up army AI. In the meantime, Biden retains calling for Congress to go binding AI rules. It’s no shock, then, that Silicon Valley’s billionaires have backed Trump.
