U.S. senators and tech executives seemed open Tuesday to regulating the exploding field of artificial intelligence, but after Congress has failed to act when presented with multiple opportunities to rein in technology, don’t bet on it actually happening.
At a Senate Judiciary subcommittee hearing on Tuesday called “Oversight for AI,” senators heard testimony from Sam Altman, the chief executive of OpenAI, which created ChatGPT; Christina Montgomery, privacy and trust officer at IBM Corp.
IBM,
; and Gary Marcus, an NYU professor emeritus. All of the witnesses agreed that some form of regulation should be forthcoming, but they disagreed on what form.
Altman suggested the government have licensing, oversight and standards-setting mechanisms, while also raising fears that AI could cause “significant harm to the world.” Montgomery suggested establishing rules to govern the deployment of AI in specific use-cases without regulating the technology itself, and Marcus suggested a neutral global entity similar to nuclear-research regulator CERN that would assess systems before they are released to the public in a clinical-trial-like evaluation.
When Sen. Cory Booker, D-N.J., raised fears that there is no way to put the AI genie back in the bottle and asked Marcus for thoughts on forming a regulatory agency, Marcus said “there are other genies to be concerned about,” adding that there need to be international meetings very quickly. “I think science has to be a really important part of it,” he said.
Also see: This is how AI conquers humans: trapping us in its robotic ‘echo chamber’ with no escape
Altman suggested that licensing AI technology would likely limit the number of companies that could afford a licensing process, but said one of the benefits of that would be a smaller number of companies for regulators to keep an eye on. Marcus, the lone academic with no company ties, voiced his concern about having a small number of companies involved in the development of AI. In his prepared statement, Marcus also noted that while Open AI has suggested audits of AI technology, he did not believe the company had submitted to any audits.
“We have to stop letting them set all the rules,” Marcus said. In 2019, Open AI received $1 billion in funding from Microsoft Corp.
MSFT,
The software giant is now using ChatGPT in Bing, a move that has put it in the headlights of Alphabet Inc.’s
GOOG,
GOOGL,
Google, which rushed out its Bard competitor to ChatGPT.
Some of the senators themselves referenced the government’s inability to regulate many aspects of technology, such as data privacy and social media, with Sen. Jon Osoff, D-Ga., noting that the U.S. still does not have a national privacy law. Subcommittee Chair Sen. Richard Blumenthal, D-Conn., said that forming a new federal licensing agency would require some “real hard decision-making in how to frame the rules,” while also noting that for every success story in creating new government agencies, there are many more failures.
More from Therese: Instead of banning TikTok, Congress should do its actual job
Congress has proved capable at hauling tech executives to Capitol Hill for these type of sessions, but has proved frustratingly incapable of the hard decision-making or reaching agreements across the aisle that goes into making actual laws. While it is a sign of progress that senators at least seem to realize their failures of recent years, it is time for them to act instead of ceding control of tech regulation to Europe.
As more scary AI scenarios emerge, it has become imperative that AI development face some sort of regulation, whether it be ethics reviews, alerting consumers they are interacting with AI or whether an AI app is high risk. But it already looks as though the European Union may be again ahead of the U.S., with the proposed AI Act, which is headed to the European Parliament before heading to the European Commission.
Marcus warned in his opening statement that “We have built machines that are like bulls in a china shop — powerful, reckless and difficult to control.” The U.S. might be able to create some guardrails for this technology if Congress can get its act together, but given what has happened in the past decade, it is much more likely we will see nothing beyond talk, as Europe takes the lead on actual action.
Read the full article here