The AI Daily Brief: Artificial Intelligence News and Analysis - Sam Altman and Greg Brockman COMING BACK to OpenAI?!

Episode Date: November 19, 2023

Act 3 of the Dumbest Palace Coup Ever, as employees and investors revolt and the board appears forced to capitulate.  ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, Act 3 of the dumbest palace coup ever, as it appears Greg Brockman and Sam Altman might be on their way back to Open AI. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.net network for more information about our YouTube channel, our Discord, and our newsletter. All right, friends, welcome back to the AI breakdown. Today, of course, we are on part three of our emergency episodes, which have happened ever since Sam Altman was announced to be fired. from OpenAI on Friday. As with the other episodes in this series, I am doing these without editors,

without my normal team, so they are a little bit more off the cuff. They are a little bit more not so refined, let's say, because they are happening fast and in real time and with as little editing as possible to just get them out to you. So I'm calling this act three of the dumbest palace coup ever, and act one, of course, was the firing.

Act two, which we started to talk about yesterday, was the backlash. Act three is the return. Mark? Where did we leave off yesterday? A few things to remember. First of all, we had learned that this wasn't some big infraction from Sam. At first, when this all came out, it seemed like the only plausible explanation was that Sam had done something extraordinarily wrong, right? Some sex scandal or some state secret level challenge. Just the way that the board worded this, it made it seem extremely dramatic. However, as time went on, it started to seem more like it was just a battle

about the direction of the company in which chief scientist Ilya Sutskhaver, who is the one here on the far, if you are watching the video, who doesn't have an X over his chest, didn't like how the company was proceeding. He didn't like how his role was coming. And indeed, it's even unclear to what extent this is the EA versus EACC, the effective altruist versus the effective accelerationalist, AI safety versus accelerationism battle versus just a garden variety power struggle within a big, important company. Now, obviously right now it is being interpreted as an AI safety related battle between the effective altruist align section and the accelerations, but it's not at all clear to me that's the case. However, it doesn't

really matter right now. What does matter functionally is that Silicon Valley certainly was taking this as war. It really touched a nerve to have this company, which many had seen bringing the mojo back to Silicon Valley, bringing the technology excitement back to San Francisco. For that company to be targeted in this way or for the founder of that company to be pushed out in this way was just beyond the pale for that entire community. And that's been the color of the conversation ever since. Now, the other thing that we had going on yesterday as we were digging into this and starting to see the backlash is the rumble of angry investors back behind the scenes, right? There were a lot of indications that whatever public statements they were making, Microsoft and particularly Satya Nadella, were very upset about this. The VCs who had backed Open AI were not quiet about it at all, being blindsided by this, having found out just minutes before or in some cases after it actually happened.

And so there was a lot of burbling and bubbling that was going on at yesterday around mid-afternoon East Coast time. And at that point, my base case had started to switch. When this all went down, it seemed very clear that there would be a hit to Open AI's value. There's just no way that there wasn't going to be. Even if it ended up being justified, it was going to create questions around how well they were going to be able to deliver on their mission, how well they were going to be able to compete. it seemed to potentially create an opening for other competitors like Google. And so I had initially, when I was working under the assumption that Sam must have done something really wrong, I was initially thinking that maybe that would result in something like a reduced valuation from the employee tender share offer that they had been working on,

more like $40 to $50 million valuation instead of $80 to $90 billion valuation, or, and or I guess, a full sort of acquisition for Microsoft being a likely next step, right? maybe they were going to come in and consolidate. Now, as it became clear that this was not, in fact, some egregious action from Sam Altman, or at least the board has chosen not to make that argument in the time that they could have as the whole world has exploded after this announcement. But instead, again, just a disagreement about the nature of the or the direction of the company, my base case shifted pretty dramatically.

My base case, in fact, shifted to Sam and Greg coming back in short order and the four remaining members of the board, current remaining members of the board, being kicked out, being forced to resign basically at gunpoint based on the reaction to this. So that had become my base case. Now, a couple other things just to look at what we learned yesterday before we get into sort of the big shift that happened last night. To the extent that there was any indications of things that Sam had been doing, quote unquote, wrong, one of the things that there seemed to be acrimony around was the idea that he was out raising money for additional companies are potentially exploring raising money for additional companies, including an AI chip

rival to go after the space inhabited by Nvidia, as well as the conversations around him and former Apple designer Johnny I of designing some sort of new AI phone or AI type native device. That was part of what was suggested by Bloomberg and some other reportage as a big area of consternation. Maybe Sam was out trading against the open AI name. A little sort of flimsy to me, even if it was happening, but that was certainly part. of the conversation and the reporting. Now, other things that were going on, again, during the day before the big dramatic thing happened, which we'll get into in just a moment, was that it was very clear that this sort of this $86 billion price share sale that had been in the works

was no longer looking as likely to close. And those employees who had been planning to sell their shares in that were potentially looking at not being able to or doing so at a deeply discounted rate, at least relative to that big sale price. So again, this was an indication of how the market was treating this news, broadly speaking. Now, on the flip side, it was basically no time before venture capitalists started lining up to fund whatever the heck Sam Altman and Greg Brockman wanted to do next. And within hours of this whole thing going down, sources like the information were reporting that, yes, indeed they were planning on launching a new AI venture, although it wasn't clear whether it would be a direct competitor to Open AI or a pursuer of one of

these other opportunities that had been one of the sources of consternation initially as well. The final thing that happened that we'll talk about before we get into sort of this big shift is that the open opportunity, the open season that was declared on Open AI staff was really taken up by a ton of different competitors. Sources suggest that Google's deep mind saw a significant uptick in new applications from Open AI employees. There were other startups like Coher and Adept that were aggressively trying to do. trying to poach and pursue those employees. And again, just another indication of the extent to which the market was interpreting Open AI

as in some amount of free fall and chaos and there being a moment for other players to rush in and take advantage of those shifts. Now, the thing that came clear pretty quickly is what I was saying in my base case, that the amount of money on the line here, the depth of the bet that has been made on this company, suggested that it was unlikely that they were going to be able to stand on principle for much longer. And indeed, by around 5 p.m., 6 p.m. Eastern time yesterday, the Verge and others were reporting that the board was already in discussions for Sam to come back as CEO. So in point of fact, there were actually two things going on simultaneously, probably more than that.

But the two that I'm referring to are on the one hand, you had this back and forth conversation between Sam Altman and presumably Greg Brockman and the Open AI board with a lot of pressure coming in from all sides from investors as well. And then you also had an employee-led revolt where the people who remained at the company were trying to exert their own will on this process. Now, we had seen on Friday night after the announcement a number of key staff resign and make a lot. a statement in that way, and it was anticipated that that might expand and spread over the course of the next day or so. And certainly as that announcement was coming out from the verge that the board was talking with Sam about a return, it also broke that OpenAI staff had set a deadline of 5 p.m. Pacific or 8 p.m. Eastern for all board members to resign and for Sam and Greg to be back, or else they would resign. Now, apparently, initially the board had agreed to that, but they started to waffle and about an

hour after the deadline, there wasn't any real conclusion. And so this is where we started to see a mass action on X slash Twitter. It started first with people noticing individual members of OpenAI's team had done things like change their bios. Rowan Chung writes, breaking Logan K, official Logan K of OpenAI has removed developer relations at OpenAI from his bio. Did OpenAI staff just quit and join Sam Altman in his new company? But then it got even more dramatic. All of a sudden, Sam Altman tweeted, I love the OpenAI team so much. And dozens, if not hundreds of OpenAI team members reposted it with heart emojis. Basically, that was the standard message.

You had Greg Brockman. You had even Miramirati, the CTO who had replaced Sam as interim CEO, responding with a heart. And it wasn't clear to people watching from the outside if this was, one, a just natural sort of spontaneous expression of support from the team, or if it was, too, an indication that there had been a shift in the board stance and that they weren't going to comply with employee demands, and this was a way to exert public pressure on that whole process. In any case, by around 10.30 p.m. Eastern Time, we got more information leaks from internal open AI conversations in which Chief Strategy Officer Jason Kwan had told members of the team in a memo on Saturday night that, quote,

we are still working towards a resolution and we remain optimistic. By resolution, we mean bringing back Sam, Greg, Jacob, Simon, Alexander, and many other colleagues, and remaining the place where people who want to work on AGI, research, safety, products, and policy can do their best work. Now, as this reporting was happening, it was also coming out that all day reporters had seen senior employees from OpenAI coming to Altman's house in San Francisco. going back and forth and presumably trying to have a hand in how to figure out the resolution to all of this madness. Again, in that memo, Kwan said,

A ton of people have worked incredibly hard and sacrificed much to make Open AI what it is. This has never been more clear than in the last 48 hours, seeing all the ways in which the people who work here have done what they could to save the company, from passing along messages, supporting each other, coming into work over the weekend, and yes, even leaving in protest to send a message. Let's make sure it lasts. Now, of course, one of the fault lines is that if Sam were to come back, he was certainly not going to do it with this board still in place. Wrote the information. Sam's return would invite a new bout of upheaval.

The members of the board would have to be replaced, according to a person familiar with the discussions. Alternatively, Altman would continue with plans to launch a new venture. This is something that a lot of people noticed just the incredible strategic blunder of these folks if they thought that they were going to get away with this. Will Minitas tweeted, insane that if Sam returns he gains one, total full for-profit structure because non-profit is too dangerous after this, two, a totally aligned board with no legacy EA baggage, three, organization free of weirdo-dumers. Imagine trying to destroy someone and instead giving them complete power. Will added, to be clear, I don't think this was orchestrated, I just think these people suck at coups.

Echoing that idea that any change like this was going to come with governance shifts was Marissa Mayer, CEO of Yahoo saying, I've read that Sam Altman is asking for governance changes, which is completely appropriate. But it shouldn't just be the board resigning. I'm fundamentally an AI optimist and realizing the structural flaws in Open AI's governance has given me serious pause.

Then, of course, there was the extension of this EACC insurgency and this culture war that had come up in such brutal fashion between the effective altruists and the EACC's, which is really coming to a head and I think will be the subject of a conversation later this week, as I believe that it has big implications for the AI safety space. Brian Armstrong, the CEO of Coinbase, wrote,

If this is really some EA D-Sel AI safety coup at OpenAI, the board just torched $80 billion of value, destroyed a shining star of American capitalism, and will be sued to high heaven by investors. Every talented employee at OpenAI should quit and join Sam and Greg's new thing if they make one. This time skipped the woke nonprofit board, eject the DECLs slash EAs, maintain founder control,

avoid nonsensical regulation, and just build. Accelerate progress. You're building something good for the world. Don't let anyone make you feel guilty for it and try to capture it for their own motives. Beth Jazo's quote tweeted a photo of the Effective Altruist Forum where someone pointed out that EA was not coming out well in this whole culture fight. The forum post wrote, I voiced concerns before that EA is unaware that it can be drawn into one-way fight sometimes and this feels like another such moment. The Silicon Valley Tech Twitter scene has exploded over this and so far EA is not coming out well in their eyes from what I can see. I think the days of EACC being a me movement are rapidly drawing to a close, and EA might find itself in a hostile atmosphere in what used to be one of the most EA-friendly places in the world.

Beth Jzos wrote, good, SF is EACC town now. EA's no longer welcome after destroying $100 billion in value overnight. Now, as of Sunday when I'm recording this at around 11.30 a.m. Pacific time, 2.30 p.m. Eastern time, we have not gotten any further updates. There hasn't been any sort of formal transition of power back to Sam. But what we did get just recently was a tweet from one of the voices who had been quiet in this so far, Andrei, Twitter user X solely wrote, Carpathy's silence is interesting. He's my favorite one out of everyone at OpenAI, followed by Greg Brockman. I wonder why he still did not comment on current events.

Andre responded, I just don't have anything too remarkable to add right now. I like and respect Sam, and I think so does the majority of Open AI. The board had a chance to explain their drastic actions, and they did not take it. So there is nothing to go on except exactly what it looks like. And that is certainly where I find myself as well. There is nothing to go on except exactly what it looks like. Increasingly, it looks like just an unbelievably naive, ill-considered, and ultimately radically ineffectual coup attempt that accomplished none of its goals, whatever those goals might have been.

In fact, it seems increasingly clear to me that to the extent that this was about AI safety and the speed of progress, it may go down as not only not being effective, in whatever its goals were, but unintendedly accomplishing the exact opposite. Whether it's because Sam Altman is now free to go start his own company, to go at the speed that he wants with the full backing of the entire technology establishment, or because in his return, there is an absolute culling of the mindset which would slow down progress. In either case, the result is the same. Faster AGI, with fewer guardrails and less consideration for the consequences of the action. I'm sure I will be back to

tomorrow with another major update. And my guess is the update will say, Sam and Greg are back at OpenAI. Until next time, peace.