Two opportunities to work with me in 2025
You could get paid to write for Understanding AI for nine months.
Until now, Understanding AI has been a one-person operation, supplemented by occasional guest posts. But there’s way too much happening in the world of AI for one person to cover. So I’m looking for a couple of people who can help me broaden the newsletter’s coverage:
I’m participating in a fellowship program that could pay someone to write for Understanding AI. The one-year program comes with a $50,000 stipend.
I’m looking for collaborators to help me test LLMs more rigorously—perhaps a researcher who is already developing LLM benchmarks but would like to reach a larger audience.
If either of those opportunities sounds interesting to you, please read on.
The Tarbell Fellowship
The Tarbell Center for AI Journalism is a non-profit that financially supports journalism focused on AI. Their flagship program is the Tarbell Fellowship, a one-year program for early-career journalists interested in covering artificial intelligence. After three months of training and networking with their peers (June 2025 through August 2025), Tarbell fellows are placed in newsrooms for nine months (September 2025 through May 2026) to write stories about AI. Last year, Tarbell placed fellows with Time, the Information, the MIT Technology Review, and other publications. Fellows receive a $50,000 stipend.
Understanding AI will be a host publication in 2025, and I’m hoping I have readers who would be a perfect fit. Prior journalism experience is preferred but not required. I’d especially like to hear from folks with a technical background who are interested in transitioning to a career in journalism. I made a similar transition years ago and have experience writing technically-grounded explainers that are valuable to broad audiences.
I strongly prefer people who are willing to relocate to Washington DC (where I’m based) for the duration of the fellowship. But other publications are located in other cities and some may be open to remote work. So I encourage you to apply regardless of your geographical constraints.
Tarbell is holding a virtual information session for prospective fellows next week—on Wednesday, January 15 at 1 PM Eastern Time. To participate you just need to click on this Google Meet link. I’m also happy to answer questions directly—email me at tim@understandingai.org.
You can click here to apply for the fellowship. The application deadline is February 28.
Benchmarking collaborators
Every time one of the leading labs releases a new model, I use a battery of tests to try to evaluate its performance relative to other models. While I’m proud of these stories overall, I have to admit that my testing is fairly ad hoc and impressionistic. Moreover, as the performance of these models has improved, it has gotten more difficult to come up with problems that are easy for readers to understand but hard for models to solve.
So I would love to work with someone who can help me test new models in a more rigorous and systematic way. I can’t afford to hire someone to work on this, but maybe there are folks out there who would like to collaborate. For example, maybe you’re a researcher who has been developing benchmarks and would like to see them reach a wider audience. Maybe you have a startup that could help me build tests like this efficiently. Maybe you’re an undergraduate looking for a topic for your senior thesis.
Regardless. I’m interested in talking to anyone who has been exploring ways to better understand the capabilities and limits of LLMs. If you’d like to talk to me, just respond to this email with a brief introduction.