The UChicago AI safety group had humble beginnings. One day in 2022, after a dinner hosted by the school's EA group, a student was asking if anyone would be interested in attending the inaugural UChicago AI Alignment Research Group meeting. One other student and I said yes, and three or four more met up with us later.
We walked across campus to the Woodlawn dorms, the newest building on campus but of the lowest quality. Many of the building's walls were concrete. If you are a sufficiently nerdy person, you would know this is great news because you can write on concrete with chalk, so everything vertical becomes a blackboard. We decided to do our meetings in the stairwells for privacy and lots of large open walls to write.
There was no food, funding, or mentorship. We weren't a registered student organization, so we didn't have the ability to book rooms or get support from the University. There was no point in networking because nobody was important and nobody knew anyone important.
This was a place and moment where the only reason one would join an AI safety group is because they wanted humanity to survive the Singularity. Even though it seemed unlikely we could do anything, there was a sense that we would try.
A few months prior, some organizers ran an EA Midwest retreat, funded by a small grant from OpenPhil (or maybe another EA funder, I'm not 100% certain). Students came from UChicago, Northwestern, UIUC, UW Madison, University of Michigan, along with a collection of other EAs living in the general Midwest area. There was a lot of talk about AI safety during this retreat. Many of the people in attendance went on to be successful in AI safety (for privacy, I won't use names here, but these are names you may recognize).
I wasn't an organizer, but it seems there was a vibe among the organizers that they shouldn't spend money just because they could. The attitude was that spending when not strictly necessary would create the wrong tone for the event and that OpenPhil could use unspent funds on other effective grants. Operationalizing this philosophy, one afternoon, instead of ordering food, they decided to cook pasta. The organizers went to the grocery store, bought spaghetti and vegan meat. They boiled the water. Cooked the pasta. Washed the dishes.
I can imagine an organizer thinking to themselves, "I can't fucking believe that I'm sitting here washing dishes when we have THOUSANDS OF DOLLARS that OpenPhil gave us to spend ON FOOD and instead I'm spending my valuable time cooking pasta and washing dishes so we can effectively donate back money THAT OPENPHIL ALREADY DONATED TO US."
OpenPhil probably would have agreed in the sense that spending a few hundred dollars to order food and save everyone some time and stress seems like a reasonable trade-off. So maybe this is a good argument. But we could also view it as a good selection effect. Those who were unwilling to wash dishes would have had a bad experience and therefore would have been less likely to be involved in the future. For those that care primarily about money and status, the experience would be especially off-putting. Regardless of how conscious the effort was, the organizers were engineering a culture to preserve a specific set of values.
I can't imagine going to an AI safety event today where organizers cook the food themselves. Likewise, AI safety student groups today aren't meeting in hallways. They have actual funding (in some cases 10k+ per year) and many have their own office space.
I'm not claiming this is bad, especially in the university group case, which is still pretty scrappy. But it's also a trade-off. It's hard to get away from the fact that the nicer you make retreats, the higher the salaries you pay people, and the more prestige your university group has, the less you select for altruism and for those willing to make personal sacrifices.
The alignment community now is well funded, and there isn't as much of a sense, even at universities, that we, as a community, need to be frugal. Internships at AI safety orgs can be $100 an hour, even for undergrad-aged employees. A competitive stipend for a summer research fellowship is the equivalent of $1k per week but stretches significantly higher. Hilariously, Dario Amodei recently lamented that he suspects that people are taking jobs at Anthropic for the money and not because they believe in the mission. In Anthropic's case, it's obvious, but of course it's more subtle most of the time.
"Salad days" is a term coined by Shakespeare in the play Antony and Cleopatra. Cleopatra explains away her relationship with Caesar, saying she was young and naive and living in her "salad days." The phrase has evolved a lot over time and, like many metaphors, it's hard to tell what it means anymore. The interpretation I like is that "salad days" is the phase of life when you eat salad by necessity because you can't afford meat or wine. And you are happy to live like that because you are still green and cold-blooded. When subcultures have "salad days," they function as a filter for integrity. It's analogous to "eating your vegetables," but instead of building a healthy body, you are growing a healthy culture where people are there for the right reasons.
I'm writing this in an attempt to document something important from the past: that before there was funding, high-paying internships and a vibrant social community with lots of cool well-connected people, there were salad days. There was a time when the AI alignment movement was green and cold-blooded. I think it was a moment worth remembering.
The UChicago AI safety group had humble beginnings. One day in 2022, after a dinner hosted by the school's EA group, a student was asking if anyone would be interested in attending the inaugural UChicago AI Alignment Research Group meeting. One other student and I said yes, and three or four more met up with us later.
We walked across campus to the Woodlawn dorms, the newest building on campus but of the lowest quality. Many of the building's walls were concrete. If you are a sufficiently nerdy person, you would know this is great news because you can write on concrete with chalk, so everything vertical becomes a blackboard. We decided to do our meetings in the stairwells for privacy and lots of large open walls to write.
There was no food, funding, or mentorship. We weren't a registered student organization, so we didn't have the ability to book rooms or get support from the University. There was no point in networking because nobody was important and nobody knew anyone important.
This was a place and moment where the only reason one would join an AI safety group is because they wanted humanity to survive the Singularity. Even though it seemed unlikely we could do anything, there was a sense that we would try.
A few months prior, some organizers ran an EA Midwest retreat, funded by a small grant from OpenPhil (or maybe another EA funder, I'm not 100% certain). Students came from UChicago, Northwestern, UIUC, UW Madison, University of Michigan, along with a collection of other EAs living in the general Midwest area. There was a lot of talk about AI safety during this retreat. Many of the people in attendance went on to be successful in AI safety (for privacy, I won't use names here, but these are names you may recognize).
I wasn't an organizer, but it seems there was a vibe among the organizers that they shouldn't spend money just because they could. The attitude was that spending when not strictly necessary would create the wrong tone for the event and that OpenPhil could use unspent funds on other effective grants. Operationalizing this philosophy, one afternoon, instead of ordering food, they decided to cook pasta. The organizers went to the grocery store, bought spaghetti and vegan meat. They boiled the water. Cooked the pasta. Washed the dishes.
I can imagine an organizer thinking to themselves, "I can't fucking believe that I'm sitting here washing dishes when we have THOUSANDS OF DOLLARS that OpenPhil gave us to spend ON FOOD and instead I'm spending my valuable time cooking pasta and washing dishes so we can effectively donate back money THAT OPENPHIL ALREADY DONATED TO US."
OpenPhil probably would have agreed in the sense that spending a few hundred dollars to order food and save everyone some time and stress seems like a reasonable trade-off. So maybe this is a good argument. But we could also view it as a good selection effect. Those who were unwilling to wash dishes would have had a bad experience and therefore would have been less likely to be involved in the future. For those that care primarily about money and status, the experience would be especially off-putting. Regardless of how conscious the effort was, the organizers were engineering a culture to preserve a specific set of values.
I can't imagine going to an AI safety event today where organizers cook the food themselves. Likewise, AI safety student groups today aren't meeting in hallways. They have actual funding (in some cases 10k+ per year) and many have their own office space.
I'm not claiming this is bad, especially in the university group case, which is still pretty scrappy. But it's also a trade-off. It's hard to get away from the fact that the nicer you make retreats, the higher the salaries you pay people, and the more prestige your university group has, the less you select for altruism and for those willing to make personal sacrifices.
The alignment community now is well funded, and there isn't as much of a sense, even at universities, that we, as a community, need to be frugal. Internships at AI safety orgs can be $100 an hour, even for undergrad-aged employees. A competitive stipend for a summer research fellowship is the equivalent of $1k per week but stretches significantly higher. Hilariously, Dario Amodei recently lamented that he suspects that people are taking jobs at Anthropic for the money and not because they believe in the mission. In Anthropic's case, it's obvious, but of course it's more subtle most of the time.
"Salad days" is a term coined by Shakespeare in the play Antony and Cleopatra. Cleopatra explains away her relationship with Caesar, saying she was young and naive and living in her "salad days." The phrase has evolved a lot over time and, like many metaphors, it's hard to tell what it means anymore. The interpretation I like is that "salad days" is the phase of life when you eat salad by necessity because you can't afford meat or wine. And you are happy to live like that because you are still green and cold-blooded. When subcultures have "salad days," they function as a filter for integrity. It's analogous to "eating your vegetables," but instead of building a healthy body, you are growing a healthy culture where people are there for the right reasons.
I'm writing this in an attempt to document something important from the past: that before there was funding, high-paying internships and a vibrant social community with lots of cool well-connected people, there were salad days. There was a time when the AI alignment movement was green and cold-blooded. I think it was a moment worth remembering.