A rift appears to have opened at OpenAI as top execs quit and safety concerns mount
ScarJo’s voice ain’t the only drama… More turmoil is brewing at OpenAI after two of the ChatGPT maker’s top safety execs resigned. OpenAI chief scientist and cofounder Ilya Sutskever announced his departure on X last week, and research leader Jan Leike followed suit within days. Both led the company’s “superalignment team,” which was dedicated to preventing smarter-than-human AI “from going rogue.” OpenAI, which thinks that superintelligent AI could arrive this decade, said the risks could include “disempowerment of humanity or even human extinction.”
On that note… The superalignment team itself has been dissolved. OpenAI said its work is being redistributed, but it’s unclear who (if anyone) will lead on safety.
Beef’s been brewing: Last year, a boardroom revolt led to the surprise (short-lived) ouster of boss Sam Altman. The board was reportedly concerned that Altman was moving too fast.
The safety rift… The recent exec departures have raised concerns over OpenAI’s internal safety practices. After Altman’s temporary firing, a split appears to have emerged within the company: those who think it’s moving irresponsibly fast and those (like Altman) who think it should be doing more to grow its biz. In his resignation post, Leike said he’d reached a “breaking point” with OpenAI leadership over the company’s priorities.
Leike wrote: "safety culture and processes have taken a backseat to shiny products." Perhaps products like the Scarlett Johansson-sounding voice assistant, though that may be the least of our concerns…
Another safety employee who left OpenAI said he’d lost confidence that the company would behave responsibly if it ever reached its goal of creating AI that can outperform humans (aka: artificial general intelligence or AGI).
Capitalism doesn’t like to go slow… OpenAI was founded as a nonprofit in 2015 with a mission to build safe AGI “for the benefit of humanity.” Three years later, OpenAI announced a “capped profit” structure (though it’s still governed by a nonprofit). But after a $10B investment from Microsoft and a viral product, it’s moving a lot more like a regular business. As shown in its splashy demo this month, OpenAI has been racing to release new tech to one up competitors like Google and was quick to monetize its products through CGPT subscriptions, licensing deals, and the GPT Store.