For OpenAI’s CEO, the rules don’t apply
Since my last newsletter, a shakeup at OpenAI somehow caused Sam Altman to be fired, hired by Microsoft, and then re-hired to his original post in less than a week’s time. Meet the new boss, literally the same as the old boss.
There are still a lot of unknowns about what went down behind closed doors, but the consensus is that OpenAI’s original board fired Altman because they thought he was building risky, potentially harmful tech in the pursuit of major profits. I’ve seen other media calling it a “failed coup”, which is the wrong way to understand what happened. Under the unique setup at OpenAI — which pledges to “build artificial general intelligence (AGI) that is safe and benefits all of humanity” — it is the board’s job to hold the CEO accountable not to investors or even to its employees, but rather to “all of humanity.” The board (alongside some current and former staff) felt Altman wasn’t holding up his end of the deal, so they did their job and showed him the door.
This was no coup. But it did ultimately fail. Even though Altman was part of the team that created this accountability structure, its rules apparently no longer applied to him. As soon as he left, his staff apparently threatened to quit en masse. Powerful people intervened and the old boss was back at the helm in time for Thanksgiving dinner.
Now, OpenAI’s board is more pale, male and I dare say stale than it was two weeks ago. And Altman’s major detractors — Helen Toner, an AI safety researcher and strategy lead at Georgetown University’s Center for Security and Emerging Technology, and Tasha McCauley, a scientist at the RAND Corporation — have been shown the door. Both brought expertise that lent legitimacy to the company’s claims of prioritizing ethics and benefiting “all of humanity.” You know, women’s work.