On November 20th, The Information reported the following (https://www.theinformation.com/articles/openai-approached-anthropic-about-merger):
> OpenAI’s board of directors approached Dario Amodei, the co-founder and CEO of rival large-language model developer Anthropic, about a potential merger of the two companies, said a person with direct knowledge. The approach came after OpenAI’s board had fired CEO Sam Altman on Friday and was part of an effort by OpenAI to persuade Amodei to replace Altman as CEO, the person said.
This was then disputed by Will Knight, a Wired reporter, who tweeted (https://twitter.com/willknight/status/1726793735143621058)
> I have reason to believe that the OpenAI Anthropic merger story is a complete fabrication
On Feb 1st 2024, will I think The Information's reporting, as represented by the quote placed in this article, is accurate? I promise I will try to be reasonable.
Since this is about my beliefs, I will not bet in this market.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ249 | |
| 2 | Ṁ168 | |
| 3 | Ṁ153 | |
| 4 | Ṁ107 | |
| 5 | Ṁ89 |
People are also trading
In case of interest, Wired just reported that: "OpenAI board members even contacted Amodei and asked if he would consider merging the companies, with him as CEO. Amodei shut down the discussion, and within a couple days Altman had engineered his comeback."
https://www.wired.com/story/anthropic-benevolent-artificial-intelligence/
@DanielFilan Some off-the-cuff thoughts:
- without any specific reporting aside from the OpenAI fiasco more generally, my prior would be like, 20-60% this happened? If I'm Toner, and the entire staff is threatening to walk to Microsoft, and letting the company fail could be consistent with mission, feeling out Anthropic seems like something worth doing to potentially strengthen one's BATNA. Big distinction between 'quick phone call' from a board member where the word 'merger' was said or implied (I'm assuming this counts?) vs any formal or extended discussions (this seems multiple times less likely).
- Information article would therefore, on reasonable Bayesian updating, suggest it was quite likely this happened.
- Will Knight tweet, if it stood, would largely negate the positive update for me - Wired seems credible enough on this topic?
- But the deletion/clarification of the tweet pretty much negates this negation for me (lol). On the one hand, maybe it was just deleted out of prudence. On the other hand, the the quick deletion also pattern matches to 'one of my sources said 'uh, this actually happened so might want to walk that back'. So I think the sum of the Will Knight tweet saga could be a positive or negative bayesian update. I think a reasonable person could update merger discussion down from their priors by up to a quarter or so considering the totality of Will's tweets, but not more.
Gun to my head, considering binary YES/NO, the information article is enough to tip things into Yes(i.e. more likely than not) from a 40% prior or so. But given how little information we have, and reasonably differing priors, my gut favors a market/50%/N/A type resolution.
I'm a little biased as a yes holder although like to think my price for my principles is higher than 90 mana haha
@DanielFilan I think that if this was true, it would be leaked by a lot more than just one source, as it would have been a very juicy piece of information.
If it's true, and someone with knowledge of it knows and was willing to talk to The Information about it, then it's probably not a top-secret conversation that is known only by the board of directors and Anthropic's CEO (as they're unlikely to be the source of the leak themselves), and I'd expect other sources to be willing to confirm the event.
If it was fabricated by The Information or one of its sources, then the best any other person close to the situation (aside from the board and Anthropic's CEO) can say is "I don't know, but I doubt it happened", because there's always a chance of an extremely secret meeting that happened behind closed doors in the heat of the moment. This would leave a journalist unable to prove or disprove the claim.
My best guess is that it didn't happen, that some source with a lot of knowledge about the ousting, but not full knowledge, knee-jerk reacted and told Will Knight that this is nonsense, then reconsidered and realized that they can't rule out a super-secret meeting they don't know about, and contacted Will Knight again to correct themselves. Will then attempted to find other sources, but no one knew for sure (otherwise he would probably have said he could confirm the rumor in his clarification tweet).
Just my best guess, though.
This NYT article does not mention the board approaching Anthropic:
https://www.nytimes.com/2023/12/09/technology/openai-altman-inside-crisis.html?unlocked_article_code=1.Ek0.l1R2.QJHek4el-CjI&smid=url-share
Will Knight has deleted the tweet linked in the description. Today he tweeted this:
I posted a tweet late last night based on some information I haven't been able to corroborate. With so many different narratives and claims flying around right now, I decided it best to delete it.
Bought some yes on this and no on the linked market to trade a little arbitrage https://manifold.markets/acertain/did-the-openai-board-approach-anthr?r=Q2Fyc3RlblN0b2x6