This is an automated rejection. No LLM generated, assisted/co-written, or edited work.
Read full explanation
AI Safety Camp is an online, part-time research program. Small teams work on a concrete AI safety project over three months, Jan-Apr, led by someone who proposed the project and is responsible for guiding it. We're now accepting applications from people who want to lead a project in the coming edition.
What we're looking for
A project with a plan: something a team of ~3-5 people can make real progress on in three months of part-time work. We welcome a wide range of approaches, technical, governance, advocacy, and otherwise. What matters is that you understand the assumptions your project rests on and can explain its theory of change. You don't need to be a senior researcher or have extensive leadership experience, and many of our best past leads didn't.
What you get
Feedback on your proposal from the organizers, and if your project is selected, we help you to recruit a team from our applicant pool (367 applicants last round). We organizers stay available throughout the program to support you as needed. In late April, we will host an online conference where teams present their results. For many leads, running a project at AISC has been the first time they led a research or fieldbuilding effort and mentored people new to the field.
Track record
We've run 11 editions over eight years: 159 projects, 789 participants. Last round we had 367 applications for 27 projects. Many alumni tell us AISC was how they got started in AI safety.
Zvi's take, from last November: “By all accounts they are the gold standard for this type of thing. Everyone says they are great, I am generally a fan of the format, I buy that this can punch way above its weight or cost. If I was going to back [a talent funnel], I’d start here.”
AI Safety Camp is an online, part-time research program. Small teams work on a concrete AI safety project over three months, Jan-Apr, led by someone who proposed the project and is responsible for guiding it. We're now accepting applications from people who want to lead a project in the coming edition.
What we're looking for
A project with a plan: something a team of ~3-5 people can make real progress on in three months of part-time work. We welcome a wide range of approaches, technical, governance, advocacy, and otherwise. What matters is that you understand the assumptions your project rests on and can explain its theory of change. You don't need to be a senior researcher or have extensive leadership experience, and many of our best past leads didn't.
What you get
Feedback on your proposal from the organizers, and if your project is selected, we help you to recruit a team from our applicant pool (367 applicants last round). We organizers stay available throughout the program to support you as needed. In late April, we will host an online conference where teams present their results. For many leads, running a project at AISC has been the first time they led a research or fieldbuilding effort and mentored people new to the field.
Track record
We've run 11 editions over eight years: 159 projects, 789 participants. Last round we had 367 applications for 27 projects. Many alumni tell us AISC was how they got started in AI safety.
Papers that came out of camp projects include Goal Misgeneralization, Representation noising effectively prevents harmful fine-tuning, Multi-Turn Jailbreaks Are Simpler Than They Seem, and Are LLM Belief Updates Consistent with Bayes' Theorem?. Projects started at camp went on to receive over $1.6M in grants, and camp conversations were the starting point for Arb Research, AI Safety Support, and AI Standards Lab. Alumni have gone on to positions at GovAI, Redwood, Conjecture, Timaeus, Apollo (including two of its founders), MATS, and DeepMind, among others.
Zvi's take, from last November: “By all accounts they are the gold standard for this type of thing. Everyone says they are great, I am generally a fan of the format, I buy that this can punch way above its weight or cost. If I was going to back [a talent funnel], I’d start here.”
Practicalities
More on our website. If you know someone with a good idea and no team, please pass this along.