Unfortunately one of AISAF’s co-founders Max Chiswick, passed away on Jan 7th, 2025 while traveling. Many people may not have directly met him unless you attended one of our Bay Area workshops as he mainly was responsible for keeping our technical infrastructure running and handling some things out in the Bay. But you all saw, in one way or another, the impact of his work on our workshops.
I’ll first share a couple of tributes and obituaries from his friends before including a message I sent to his friends and family:
https://www.thefp.com/p/my-friend-max-suzy-weiss
https://substack.com/home/post/p-154541432
https://blog.rossry.net/chisness/
One of the great tragedies of human relationships is that even our closest relationships capture only a tiny sliver of who the other person is. The way we try to grapple with that is to tell stories, imperfect, linear shadows that capture only a single path through an impossibly twisting and fractal life.
But it’s the best I can do, so I wanted to share two stories about Max:
The first is one is a particular bit of poker terminology, and the second concerns the work he and I did together on AI safety.
I suspect most of Max’s friend circle knows that he was involved in poker, but just in case you didn’t know, from an early age (high school I believe) Max was seriously playing poker, eventually turning it into a professional career for a while. I never played Max (not least of which because I’d be absolutely crushed), but we were talking about poker one day and somehow I brought up the idea of “fish,” that is novice poker players who are expected to lose and “feed” more experienced players their money.
In his usual understated way, Max calmly mentioned that he disliked calling poker novices “fish.” I was mildly taken aback by this; many poker players in my experience were quite happy to disparage lower-level players.
Max didn’t do this out of a simple sentimentality. After all when he was a professional poker player, he would sometimes make money off these kinds of players.
But Max understood the fundamental kind of corrosiveness that this rhetoric had on its speaker and its community: the “other"ing of a group of people as a set of cheap marks, robbed of their own agency. It spilled over into our AI safety work. We agreed never to have pet names for people unfamiliar with AI safety: “normies,” “muggles,” etc.
It was mark of his thoughtfulness that he was able to maintain this conviction without falling into a more maudlin, performative, and ultimately empty rhetoric that simply parroted the line “everybody matters.” Max was able to combine generosity, compassion, and intellectual honesty in a way that amplified all three.
The second story concerns November 2024, when I was lucky enough to spend nearly two weeks with Max basically joined at the hip. As was often the case with Max, AISAF (the AI Safety Awareness Foundation, our organization) was not the only thing on Max’s plate. For example, a large part of his time was taken up by AI poker camp, a project he did in conjunction with Ross Rheingans-Yoo. As such, Max was reluctant for me to introduce him as cofounder, even though he had been there from the beginning, since he wasn’t working full-time on AISAF. I think that perspective sometimes led him to underestimate the impact that he had on AISAF. This story hopefully illustrates why I thought of him as my cofounder nonetheless.
We went to a conference in Boston together for a weekend and stayed in the same hotel room together. We briefly separated in Boston before meeting up again in New York, again staying in the same studio apartment together. To be clear this was entirely Max’s generosity: Max was giving up space in his studio apartment which he had rented for himself, not something we had found together. I threw my stuff in there and slept on the only piece of large furniture: a couch.
We ate most of our meals together (almost all of which Max cooked on his trusty Foreman grill); strategized together on where to take AISAF; and worked feverishly late into the night in that tiny room over and over again, slinging code, email drafts, and communiqués back and forth.
November was a tough time for me. There was a nation-wide series of workshops that I thought was both crucial to pull off, and impossible to do so, which led to a sense of despair. I had conceived of trying to do a multi-location day of workshops on Nov 20th to coincide with the inaugural convention of international AI Safety Institutes in SF. The problem was that I conceived of this during a conference on Nov 1st, and had no real partnerships with any places outside of the Seattle area.
Max and I worked workshopped it together and worked on it in the following week. By this point Nov 20th was only a little more than 2 weeks away. I thought it would be impossible to scrape together something and coordinate with multiple locations in time to make everything work, especially so late in the calendar year. We’d need to build relationships from scratch in basically a week.
I talked over all this with Max. Max looked at me, and in his typical understated style, said without hesitation “This will be good. I think you should do it.” There’s something magical about such a vote of confidence from someone as understated and thoughtful as Max. Just a few sentences felt like a crane lifting an enormous weight off of me. Max would occasionally apologize to me that he felt he wasn’t as outwardly enthusiastic as I was and worried about the effect it had on our morale. In truth his low-key style was the key to our morale. When he said that something was good, I 100% believed him.
This kicked off a crazy week. Max completely rebuilt our homepage (the third time he had done so in as many months), scoped out potential partner organizations, and kicked ideas back and forth with me about how exactly the Nov 20th workshop would be run. I became essentially a hyperactive secretary, pushing out as many emails and calls as I could to try to get people on board, all the while trying to brush up and polish workshop materials and ML projects as quickly as I could.
In the end we ended up pulling off 6 locations in addition to two remote sessions all on the same day, which set the foundations for our upcoming nation-wide events in 2025. And I was a hair’s breadth away from calling the whole thing off, if not for those simple two sentences from Max.
So thank you Max. For everything.
Changlin