AI Leader Sees a Rare Chance in Congress for Wins on Regulation

Gary Marcus, a robotics-industry executive and a leading figure in artificial-intelligence research, hailed the US Senate for taking a rare bipartisan approach to confronting the risks from the new technology — a foundation that lawmakers must now turn into concrete proposals.

(Bloomberg) — Gary Marcus, a robotics-industry executive and a leading figure in artificial-intelligence research, hailed the US Senate for taking a rare bipartisan approach to confronting the risks from the new technology — a foundation that lawmakers must now turn into concrete proposals.

Instead of dithering for years, as Congress did with oversight of social media companies, Tuesday’s landmark Senate hearing showed an unusual degree of purpose — and speed — around finding solutions, according to Marcus, chief executive officer of Robust AI Inc. and a professor emeritus at New York University. 

“It’s only six months after ChatGPT made it big that they are trying to handle it,” Marcus said Wednesday in a Bloomberg Television interview. “I saw a lot of seriousness on the part of the senators.”

Marcus provided that assessment a day after joining industry leaders including Sam Altman, CEO of ChatGPT producer OpenAI, at a forum on AI held by a Senate Judiciary subcommittee. Altman, along with Marcus and International Business Machines Corp. privacy chief Christina Montgomery, called on US senators to heavily regulate AI technology now, while the development of the technology is fast paced.

The upbeat takeaway was especially notable against the backdrop of Washington’s deep fissures across policy and party lines, with the current debt-ceiling showdown the most notable example.

“It far exceeded my own expectations,” Marcus said. “It was really bipartisan.”

Most worrisome, Marcus said, are the risks that AI poses to democracy, especially if the technology is misused by unscrupulous actors. The right approach to regulation is to let smaller AI innovation efforts flourish while putting guardrails around the so-called large-language models that pose the greatest dangers, he said.

“There’s a lot of intersection between what different countries want, and even the companies want some alignment,” Marcus said. “There’s a lot of opportunity even though the politics are difficult.”

–With assistance from Anna Edgerton.

More stories like this are available on bloomberg.com

©2023 Bloomberg L.P.