Open AI Drama Act 3 - As You Were
Was it all a dream?
Here We Go Again
At the end of my last post on this topic, I made the earth shattering prediction that the saga hadn’t ended. Remarkably prescient of me, given that most of OpenAI had signed a letter threatening to quit if the board didn’t step down. And may I add, as a died in the wool socialist, it’s good to see the return of collective bargaining.
As expected, on 21st November, talks started to reinstate Sam Altman as CEO. Maybe slightly unexpectedly, Emmett Shear was a key part of trying to remove himself. The Verge summed it up nicely:
Clearly Emmett Shear is good at his job (apparently assisted by Brian Chesky and Bret Taylor), as by the time I got up on 22nd November, Sam was back in post, the staff were happy, and one or two must have been wondering if they were more prone to hallucinating than the models they were working on:
So it looks like the staff mostly got their way. Adam D’Angelo is on still on the board, but joined by (of course) former Salesforce Co-CEO/Chair of the Board of Twitter/Shopify board member, Bret Taylor and former Treasury Secretary Larry Summers. I didn’t really know much about the latter, but apparently he’s a controversial character so there might be more drama in OpenAI’s future!
It seems highly unlikely the board will stay with these numbers - between 7 and 10 is considered the optimal size, and 9 is quite popular.
Some the news that leaked around this time was very entertaining though, as (allegedly) :
The race condition of interim CEO Mira Murati trying to hire Sam Altman and Greg Brockford back in some way while the board were scrambling to find a new CEO
While the board might have looked like they rushed into appointing Emmett Shear, they’d already approached Anthropic CEO about the role and/or a merger
And former Github CEO Nat Friedman, Alex Wang, CEO of Scale AI
One board member was on a plane and incommunicado for several hours.
For a group of people worried about the future of humanity and AI, they sure seem fond of not thinking things through.
And Microsoft?
So where does this leave Microsoft? Probably relieved. I’m not sure Satya Nadella really wanted Sam Altman and Greg Brockman running an AI research unit, or the 700 strong OpenAI team - aside from the significant investment that would be required, I can’t imagine they’d enjoy working under the constraints that naturally come with huge corporations. They’d likely have gone to do their own thing once the dust had settled and I’d imagine their old staff would once again want to follow.
What Microsoft did want to do is show stability - keep partnering with OpenAI, make sure Sam Altman is still inside the tent, and if it does all have to fall apart, make sure everyone ends up at Microsoft.
This last point appeared to strike Marc Benioff late in the day:
It’s fair to say the responses to this showed that OpenAI staff really don’t want to work at just any old big corporation, they were only interested in the one that Sam and Greg had gone to, if that. We knew Benioff would do something though - sitting back quietly and watching a vast amount of talent go to a competitor is hardly his style.
I am 100% certain that Microsoft will want some involvement with the board - if not a seat on the board then an observer. They won’t allow themselves to be blindsided again.
I have read a few pieces online suggesting this might all have been a big publicity stunt. While that’s possible, Occam’s Razor suggests it was exactly what it looked like - an inexperienced board trying to do their best but failing to prepare for anything that came next.
Hopefully this is the last post I’ll feel the need to write on OpenAI - I much prefer when it’s a bullet point or two in the Other section, rather than the story itself!
We Don’t Know What We Aren’t Being Told
Tragedy always becomes comedy given enough time, and the final piece of news I saw on this suggested the time needed here was very short. Per Business Insider via WSJ :
So it might actually be fired for feels, which all seems a bit Yes, Minister :
