OpenAI, the maker of ChatGPT, is testing how to collect broad input on decisions affecting its artificial intelligence, its president Greg Brockman said on Monday.
At AI Forward, an event in San Francisco organized by Goldman Sachs Group Inc. and SV Angel, Brockman discussed at length how the makers of wildly popular chatbots are seeking regulation of AI globally.
An announcement he previewed is similar to Wikipedia’s model, saying people with diverse views need to unite and agree on encyclopedia entries.
“We’re not just sitting in Silicon Valley thinking we can write these rules for everybody,” he said of AI policy. “We’re starting to think about democratic decision making.”
Another idea that Brockman discussed, which OpenAI elaborated on in a blog post on Monday, is that governments around the world should coordinate to ensure that AI develops safely.
Since ChatGPT’s November 30 launch, the generative AI technology that can spin amazingly authoritative prose from text prompts has captivated the public, making the program its fastest-growing app ever . AI has also become a matter of concern for its ability to create deep fake photos and other misinformation.
In assessing the path forward for AI, Brockman looked to Wikipedia as well as elsewhere. He and OpenAI said a body like the International Atomic Energy Agency (IAEA) could ban deployments, enforce safety standards, and track computing power usage.
Another suggestion was a global agreement to limit the annual growth of Frontier AI capabilities, or a joint global project in which major governments could participate.
Sam Altman, CEO of OpenAI, last week proposed various ideas to US lawmakers to establish a safeguard for artificial intelligence, including requiring licenses to develop the most sophisticated AI models and setting up related governance. He is visiting European policy makers this week.
(Reporting by Jeffrey Dustin in San Francisco and Anna Tong in New York; Additional reporting by Belen Carreno in Madrid; Editing by Kenneth Lee)