Is Helen Toner a Useful Idiot? Behind The People Firing Sam Altman
Journalism is failing to uncover the direct connections that have resulted in the firing of Sam Altman by the board of OpenAI. Specifically, they are failing to uncover how Helen Toner’s conflicts of interest and association with incompetent people most likely mean that she is a useful idiot being manipulated by others — specifically Holden Karnofsky and her former bosses at the RAND Corporation — to hobble ChatGPT for their own benefit.
Toner’s Bosses
Karnofsky’s wife, Daniela Amodei, is the President of Anthropic, meaning that both Karnofsky and his wife have a financial interest in Anthropic succeeding above ChatGPT. Karnofsky and Toner have both been prominent figures in the effective altruism (EA) community, even if Toner is unknown outside of it. Both worked together in a mission-driven non-profit organization, where relationships tend to be close and worldviews are aligned. This is crucial when assessing Toner’s motivations and influences.
Helen Toner is an independent human being with her own autonomy, so why can I assume that, after Karnofsky left OpenAI’s board and Helen Toner replaced him, that Toner is still in any way motivated to help Karnofsky?
This is unprovable without direct access to the two, but we can get closer to inferring this than the media currently is doing.
The Media Isn’t Digging
For instance, The Daily Beast has said this about Toner:
“Toner has done multiple talks at EA conferences over the years and worked at Open Philanthropy, the charitable organization of noted EA Facebook billionaire Dustin Moscovtiz. Open Philanthropy has injected $300 million into EA aligned orgs related to AI existential risk. Toner likely directed a large part of that funding while there, since her role was to “scale-up from making $10 million in grants per year to over $200 million,” according to a profile on the Future of Humanity website.”
This doesn’t fully capture her involvement with Karnofsky. These two were very close.
Just how close isn’t really well documented — I don’t know why. in 2019 Toner gave a talk in which she praised Karnofsky. Here’s what she said verbatim:
“I want to tell a story about how I began thinking about motivation. I was working at the Open Philanthropy Project and had been there for about a year and a half. I was working on projects I really cared about. I was helping the organization operate as we scaled up our grantmaking. I was doing some exploratory work on artificial intelligence (AI) policy and strategy. I was managing a few people, doing some other projects on the side, and it was all exciting. But it was a lot. In fact, it was so much that I didn’t feel like I could do any of those things well.
I hated that feeling, so I pushed myself harder. It got to the point where — even though I cared a great deal about the work, loved my coworkers, and felt lucky to have the job — some days I just felt miserable. I wanted to run away, go live by the ocean, and never think about any of it ever again.
But eventually, I got around to doing what I should have done all along: I talked to my manager, Holden Karnofsky, about it. I’ll never forget how he reacted. He began by thanking me for bringing it up with him and saying that if it ever happened again in the future, the very first thing I should do was tell him.
He then pointed out that if my response to having too much on my plate was to put my head down and work harder, that would never [solve the problem]. If I started working evenings and weekends, I could potentially put in 25% or 50% more time. And I would be running myself down doing it, having a terrible time. I’d be risking burnout.
Alternatively, if I talked to people about it — to Holden and my other team members — and prioritized better, I would be able to spend less time on things that weren’t as important and more time on things that were. Perhaps I could work five or 10 times more effectively by being strategic about my overall workload.
More generally, during my time at the Open Philanthropy Project, I gradually developed a sense for when I was working in an unsustainable way, skirting the edge of burnout, on the verge of collapsing and dropping all of my projects. It was like drunk driving, a form of risking your own safety.”
I recognize this. I have spoken about bosses to friends in this language; this is a worker who admires, respects, and shares a common vision with their boss. I’ve been in this situation.
And at the very least, in 2019 Toner very much wanted people to think she felt this way about Karnofsky.
If Karnofsky is trying to manipulate Toner into sabotaging OpenAI to benefit Anthropic, it would not be hard at all to do so by appealing to her personal philosophy and worldview — which is staunchly incrementalist and risk averse. Again, from that same talk:
“In the back of your mind, keep track of how sustainable your activities are. I want everyone in this room to think about this right now. If the rest of your life continued as it is now, could you do it? Sometimes it’s good if the answer is no. Sometimes. It can be a great decision to sprint for a while. But be aware that you are making that choice. And the people you work with and live with should also be aware that you’re making it, in part so they can help ensure it’s temporary. And you should be convinced that when you make that choice, it is truly worth it.”
Maybe since 2019 Holden has become less conservative, but this language and framing of the decision making process strikes me as something I would expect at an HR conference at a Big Four firm (which, surprisingly, aren’t as bad as they sound!).
Toner’s work at RAND Corporation is, well, less publicly accessible, and this is a free article so I’m not about to speculate about her connections to the Federal government and how they may also be motivated to encourage her to slow AI down so they can consolidate power (via regulations, backhand data sharing, or who knows with the spooks, man).
Are They Right?
AI doomerism loves to refer to science fiction movies, but when they’re trying to be slightly intellectual the doomers will reference prominent intellectuals like Toner and Karnofsky. Or maybe more Karnofsky; the former has published little, and I cannot find much discussion of her contributions to her academic papers (I’m in the middle of a jungle somewhere far away, I can’t get her books easily either). Nonetheless, we should keep everyone accountable and ask what the qualifications of these characters are.
Karnofsky we can write off out of turn as a dunce for taking Sam Bankman-Fried’s money and opening Open Philanthropy to the PR and legal mess it is facing.
Toner, on the other hand, is still an unknown quotient. Her research is obscure and hard to find; her public speeches are light on substance and feel more like the result of sexism than the result of her hard work. What I mean by that is, I feel that Toner’s intellectual contributions may have been stymied or limited by a sexist AI economy and that’s why one finds mostly fluff from her. Alternatively, she is one of those people, much like my friend from school Kent (don’t worry, he stacks shelves for a living he won’t be reading this — well, no one will be reading this, come to think of it), who just isn’t very bright. And this not very bright person was put in a position of power to enact a shadow agenda, ruining us all.
Had the board been motivated by money, none of this would have ever happened. Pretty ironic, considering SBF, Toner, and Karnofsky are all effective altruists.
Call me crazy, but I’m starting to think EA might not actually be all that effective.
