Topline
Meta CEO Mark Zuckerberg on Tuesday argued that artificial intelligence companies have a “natural incentive” to ensure the safety of their models to avoid “significant liability,” while Nvidia’s CEO Jensen Huang argued that companies can pursue both safety and speed, as the two billionaires pushed a counterargument to more serious warnings made by Anthropic and OpenAI.
Billionaires Mark Zuckerberg and Jensen Huang pushed back against the arguement for a coordinated slowdown in advanced AI development.
AFP via Getty Images
Key Facts
In a lengthy X post, Zuckerberg laid out his argument, noting that people will not use AI agents that are “misaligned with them and…don’t do what they ask,” which gives companies a “strong natural incentive” to ensure alignment.
On the push to slow down AI capability development, the Meta CEO said “trust and alignment” are the most important capabilities, and companies that fail at this will fall behind.
Zuckerberg claimed Meta delayed shipping its new AI personal agent Muse for “several months” to ensure safety and “didn’t call for everyone else to do this before we would.”
While the Meta CEO appeared to dismiss the argument that AI companies may need to coordinate to ensure safety, he backed engaging independent evaluators and advisers—something both Anthropic and OpenAI have pushed for as well.
Nvidia CEO Jensen Huang argued against a slowdown while speaking at Salesforce’s annual Dreamforce event, noting that the choice between safety and speed was a “false” one and “You could definitely have both at the same time.”
Huang pointed to market forces and said companies should pace themselves until they are confident about “releasing something that the market would appreciate.”
Crucial Quote
Zuckerberg’s post appeared to include a shot at Anthropic and OpenAI and their push for so-called “recursive self-improvement”—where existing AI models are used to train and build more advanced newer models. “Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well.”
Contra
Anthropic CEO Dario Amodei, who issued a warning about AI safety over the weekend and urged top firms to deliberately slow the pace of frontier model development, also spoke during the Dreamforce event keynote. When asked by Salesforce CEO Marc Benioff to explain his argument, Amodei offered an analogy of a car company whose rival has had a safety incident. “It’s very tempting to attack your competitor and say, ‘These guys are unsafe we’re safe,’ but I think the more responsible way to respond to it is to say, ‘Hey, first of all, let’s look at our own record.’ We may not have had this big high-profile incident, but I’m sure we’re not perfect.” The Anthropic CEO added that this was an opportunity to “recommit to transparency…invest more in safety,” and organize the rest of the industry to set standards for everyone.
further reading
Billionaire Anthropic CEO Urges Competitors To Slow Down AI Development (Forbes)
OpenAI Isn’t Going Public This Year, Sam Altman Says (Forbes)
Leave a comment