Fill out this form to indicate interest in a grantmaking fundamentals pilot program, either as a participant (to learn about grantmaking) or as a speaker (if you’re a grantmaker)!
Someone recently left me some helpful feedback on my anonymous feedback form (by the way, you should do this too if there’s anything you’d like me to know!), and I figured it was time to answer some frequently asked questions regarding the AI Safety Grantmaking Fundamentals curriculum project I’m working on (downstream of this post).
In this issue:
How was the curriculum designed?
How is this course formatted / how are you planning on running it?
What comes after someone participates in the course?
What’s the current status of the project? Are you planning on running a pilot?
How was the curriculum designed?
Since I personally don’t have grantmaking experience, I was not directly involved in the curriculum design itself! On this project, I played more of a “convenor”-type role, which meant coming up with the idea, doing some early stage research to get a sense of what this might look like, applying for funding, and recruiting the people now working on it. One of those people is Mark Aiken, who led curriculum design:
About Mark Aiken
Mark has over twenty years’ traditional grantmaking experience in international development, including grantmaking roles with the United Nations and NGOs around the world. He has a background as a lawyer working in regulatory compliance and as a workplace trainer. As part of his transition into AI safety, he was a regrantor with grantmaking.ai. While noting the differences between AI safety grantmaking and traditional grantmaking, many of the best practices are transferable and can provide a useful scaffold for learning.
After writing the initial post that kickstarted this project, my involvement in the process looked something like:
Target audience: I had several chats with people who expressed interest in career transitioning to grantmaking, and collected data regarding their biggest blockers.
Grantmaker engagement: I also had calls with several grantmakers from CG, Longview, Macroscopic, Atlas Computing (see this post), etc to get a sense of peoples’ takes on the skills needed to do well as a grantmaker and what the most important things to include in the curriculum would be.
Note: Mark also consulted with several grantmakers and benefited from exchanges on AI safety grantmaking at a conference in Oxford.
Other: Separately, I did various things like chatting with the people behind grantmaking.ai to see if we might be able to use some of their materials as mock grant applications, chatting with someone at IAPS re: their plans for the next IAPS fellowship, and meeting with someone from Constellation to discuss the possibility of hosting a workshop + to recommend making the Astra fellowship’s grantmaking stream more legible in the application process.
I took a bunch of notes from all of these meetings and shared them with Mark, and we had some discussions about high-level course structure. Beyond that, I was mostly hands off for the module design, other than providing feedback on Mark’s drafts.
How is this course formatted / how are you planning on running it?
When you’re a grantmaker, often the most important things to have are (1) deep domain expertise in a particular sub-field of AI safety, and (2) a strong network.
A common line of pushback I’ve frequently heard on this project is something like: “BlueDot’s typical courses mostly select for people new to AI safety,” “I’m not excited about a low-commitment online course format for this.”
I think this is very fair, and I agree with all of these critiques! Some notes:
First, worth noting that this project is independent from BlueDot; they might consider running it as one of their courses depending on how our pilots go, but this is far from guaranteed.
The version of this course that I’m most excited about (99th percentile success) looks something like:
Select for people who are mid-to-high context in AI safety.
Run this as an in-person workshop somewhere like Constellation.
Invite a bunch of grantmakers as speakers, and use the curriculum to make activities (e.g. mock application review) and guide what topics speakers should be selected to cover.
From there, I think it costs ~nothing on the margin to make the curriculum freely available for anyone to review on BlueDot or another website, though I’m not sure if it would ever be “facilitated” in the same way other BlueDot courses are. (This might depend on demand + strength of applicants.)
I’m also excited about this potentially increasing the legibility of grantmaking as a career path in AIS, since most pipelines currently route people towards research roles.
What comes after someone participates in the course?
Our goal is for participants to leave with three concrete advantages:
A path into regranting. Organizations like grantmaking.ai may recruit top performers directly for regranting roles.
Grantmaking skill development. Participants will practice skills like grant evaluation, RFP writing, and field mapping, which will hopefully translate into better performance on work tests.
A portfolio of artifacts. Participants will leave with mock application evaluations, gap analyses of AI safety sub-fields, and other artifacts that they will be able to include in job applications.
What’s the current status of the project? Are you planning on running a pilot?
Mark has just finalized the first draft of the curriculum, and we’re currently scheduling follow-up calls with grantmakers to solicit feedback! We’re also in talks with several organizations and fellowships in the US and UK who may be interested in hosting pilots of the program.
Fill out this very simple form to indicate interest in participating in a pilot! (If you already expressed interest in being a fellow in the previous form we put out in this piece, no need to fill this form out again.)
Thanks for reading this far! And if I’ve gotten anything wrong here, you know where the anonymous feedback form is :)