From the second OpenAI CEO Sam Altman stepped onstage, it was clear this was not going to be a traditional interview.
Altman and his chief working officer, Brad Lightcap, stood awkwardly towards the again of the stage at a jam-packed San Francisco venue that sometimes hosts jazz concert events. Tons of of individuals stuffed steep theater-style seating on Tuesday evening to look at Kevin Roose, a columnist with The New York Instances, and Platformer’s Casey Newton document a stay episode of their in style expertise podcast, Arduous Fork.
Altman and Lightcap have been the principle occasion, however they’d walked out too early. Roose defined that he and Newton have been planning to — ideally, earlier than OpenAI’s executives have been supposed to come back out — checklist off a number of headlines that had been written about OpenAI within the weeks main as much as the occasion.
“That is extra enjoyable that we’re out right here for this,” stated Altman. Seconds later, the OpenAI CEO requested, “Are you going to speak about the place you sue us since you don’t like person privateness?”
Inside minutes of this system beginning, Altman hijacked the dialog to speak about The New York Instances lawsuit in opposition to OpenAI and its largest investor, Microsoft, wherein the writer alleges that Altman’s firm improperly used its articles to coach giant language fashions. Altman was significantly peeved a couple of current improvement within the lawsuit, wherein legal professionals representing The New York Instances requested OpenAI to retain shopper ChatGPT and API buyer knowledge.
“The New York Instances, one of many nice establishments, really, for a very long time, is taking a place that we should always must protect our customers’ logs even when they’re chatting in non-public mode, even when they’ve requested us to delete them,” stated Altman. “Nonetheless love The New York Instances, however that one we really feel strongly about.”
For a couple of minutes, OpenAI’s CEO pressed the podcasters to share their private opinions concerning the New York Instances lawsuit — they demurred, noting that as journalists whose work seems in The New York Instances, they aren’t concerned within the lawsuit.
Altman and Lightcap’s brash entrance lasted just a few minutes, and the remainder of the interview proceeded, seemingly, as deliberate. Nonetheless, the flare-up felt indicative of the inflection level Silicon Valley appears to be approaching in its relationship with the media trade.
Within the final a number of years, a number of publishers have introduced lawsuits in opposition to OpenAI, Anthropic, Google, and Meta for coaching their AI fashions on copyrighted works. At a excessive degree, these lawsuits argue that AI fashions have the potential to devalue, and even substitute, the copyrighted works produced by media establishments.
However the tides could also be handing over favor of the tech firms. Earlier this week, OpenAI competitor Anthropic acquired a significant win in its authorized battle in opposition to publishers. A federal choose dominated that Anthropic’s use of books to coach its AI fashions was authorized in some circumstances, which may have broad implications for different publishers’ lawsuits in opposition to OpenAI, Google, and Meta.
Maybe Altman and Lightcap felt emboldened by the trade win heading into their stay interview with The New York Instances journalists. However nowadays, OpenAI is keeping off threats from each course, and that grew to become clear all through the evening.
Mark Zuckerberg has not too long ago been making an attempt to recruit OpenAI’s high expertise by providing them $100 million compensation packages to hitch Meta’s AI superintelligence lab, Altman revealed weeks in the past on his brother’s podcast.
When requested whether or not the Meta CEO actually believes in superintelligent AI techniques, or if it’s only a recruiting technique, Lightcap quipped: “I feel [Zuckerberg] believes he’s superintelligent.”
Later, Roose requested Altman about OpenAI’s relationship with Microsoft, which has reportedly been pushed to a boiling level in current months because the companions negotiate a brand new contract. Whereas Microsoft was as soon as a significant accelerant to OpenAI, the 2 at the moment are competing in enterprise software program and different domains.
“In any deep partnership, there are factors of stress and we actually have these,” stated Altman. “We’re each formidable firms, so we do discover some flashpoints, however I’d count on that it’s one thing that we discover deep worth in for either side for a really very long time to come back.”
OpenAI’s management in the present day appears to spend so much of time swatting down opponents and lawsuits. Which will get in the best way of OpenAI’s skill to unravel broader points round AI, reminiscent of find out how to safely deploy extremely smart AI techniques at scale.
At one level, Newton requested OpenAI’s leaders how they have been serious about current tales of mentally unstable folks utilizing ChatGPT to traverse harmful rabbit holes, together with to debate conspiracy theories or suicide with the chatbot.
Altman stated OpenAI takes many steps to stop these conversations, reminiscent of by reducing them off early, or directing customers to skilled companies the place they will get assist.
“We don’t wish to slide into the errors that I feel the earlier era of tech firms made by not reacting shortly sufficient,” stated Altman. To a follow-up query, the OpenAI CEO added, “Nonetheless, to customers which are in a fragile sufficient psychological place, which are on the sting of a psychotic break, we haven’t but discovered how a warning will get by means of.”
