Sam Altman wows lawmakers at closed AI dinner: 'Fantastic…forthcoming'
- OpenAI CEO Sam Altman spoke to an engaged crowd of around 60 lawmakers at a dinner Monday about the advanced artificial technology his company produces and the challenges of regulating it.
- The dinner discussion comes at a peak moment for AI, which has thoroughly captured Congress' fascination.
- Altman will testify before the Senate Judiciary subcommittee on privacy and technology Tuesday, alongside an IBM executive and NYU professor emeritus.
OpenAI CEO Sam Altman spoke to an engaged crowd of about 60 lawmakers at a dinner Monday about the advanced artificial technology his company produces and the challenges of regulating it.
The wide-ranging discussion that lasted about two hours came ahead of Altman's first time testifying before Congress at a Senate Judiciary subcommittee on privacy and technology hearing on Tuesday. IBM Chief Privacy and Trust Officer Christina Montgomery and New York University Professor Emeritus Gary Marcus will also testify at the hearing, which is focused on AI oversight.
The dinner discussion comes at a peak moment for AI, which has thoroughly captured Congress' fascination. On Tuesday, at the same time as the meeting where Altman will testify, the Senate Homeland Security and Governmental Affairs Committee is hosting a separate hearing on artificial intelligence in government. And on Wednesday, the House Judiciary Subcommittee on Courts, Intellectual Property and the Internet will hold yet another hearing focused on AI and copyright law.
About half a dozen members who spoke with CNBC outside of the dinner on Capitol Hill described a wide-ranging and informative discussion with Altman that spanned the many fears and hopes for opportunities that come with AI.
Altman received high praise from several members.
"I thought it was fantastic," said Rep. Ted Lieu, D-Calif., vice chair of the House Democratic Caucus who co-hosted the dinner with GOP Conference Vice Chair Mike Johnson, R-La. "It's not easy to keep members of Congress rapt for close to two hours. So Sam Altman was very informative and provided a lot of information."
"He gave fascinating demonstrations in real time," Johnson said. "I think it amazed a lot of members. And it was a standing-room-only crowd in there."
One of the demonstrations, Johnson said, was having ChatGPT, OpenAI's generative AI chatbot, write a bill dedicating a post office to Lieu. After, he had it write a speech for Johnson to deliver in introducing the bill on the House floor.
"It was a beautiful speech," Lieu quipped.
"It kind of also freaked us out," Johnson said.
Rep. Haley Stevens, D-Mich., said that despite being in her third term in Congress, she's "never been to a meeting like this," and praised Lieu and Johnson for bringing together "a total cross-section of our entire Congress to engage in a topic that is transforming our world."
Rep. Anna Eshoo, D-Calif., who co-chairs the Congressional AI Caucus, called Altman very "forthcoming" and "wonderful to have a thoughtful conversation."
"There isn't any question where he pulls back on anything," she said, adding that lawmakers had very thoughtful things to ask.
Eshoo said she had invited Altman to speak to the caucus, but that Speaker Kevin McCarthy, R-Calif., and Minority Leader Hakeem Jeffries, D-N.Y., insisted it be open to the entire chamber. Eshoo said she welcomed the opportunity.
"You have to understand something before you can accept or reject it," Eshoo said. "But then, it's like getting socks on an octopus, because it covers everything."
One of those tentacles has to do with copyright law, something House Judiciary Subcommittee on IP Chair Darrell Issa, R-Calif., has been thinking a lot about.
Issa said he's "very interested in fairly quickly providing additional guidelines for the copyright office," adding that even if entirely AI-generated content can't be covered by copyright, there needs to be guidance about when material that was created with the assistance of AI can be copyrighted.
As for Altman, Issa said that in general, "He made it clear that this can't go forward without some legislative and regulatory action, and at the same time, it would be adverse to shut down the momentum. So it's, how do you develop guardrails without sideswiping it or taking it off the road?"
Rep. Jay Obernolte, R-Calif., who has a graduate degree in artificial intelligence and sits on the congressional AI caucus, said he discussed with Altman the potential to regulate the precursors to the technology, much like is done with the raw materials needed to make nuclear weapons. Obernolte suggested this might take the form of an international registry that keeps track of which entities have enough computing power to create advanced AI.
Rep. Ro Khanna, D-Calif., whose district spans part of Silicon Valley, said Altman made two important points to members in the room.
"One is that AI is a tool, not a creature," he said. "This is something that is going to assist human beings not replace human beings. Second, that it will do tasks, not jobs. This is something that's going to help people with the jobs they have, not displace those jobs. And so I think it's been a sober conversation that's helping members understand what the tool actually does and help refute some of the hype."
Still. there are unanswered questions about the vast capabilities of AI, where Congress should step in, and OpenAI's approach to harnessing the technology. For example, some experts have critiqued the company for choosing to be less forthcoming about what went into making its latest large language model, GPT-4, something its executives have defended as an important competitive and safety move.
Khanna said the question of openness of the model is something he's discussed with Altman before, though not at Monday's dinner.
"The challenge and the value we have to contemplate is the value of having this be open source so other non-incumbents can participate," Khanna said. "But the danger of open source is they could get into the wrong hands. And there's a trade off between that."
WATCH: Can China's ChatGPT clones give it an edge over the U.S. in an A.I. arms race?