Hey folks. Ex open-source JupyterLab extension developer here.
I’m currently leading a research project at METR - a non-profit that evaluates LLM capabilities - to understand how AI speeds up open source development.
If you currently maintain an open source project with >500 stars, I’ll fund your project for $150/hour to participate in this study. Express interest to participate here.
The Details:
- Get paid $150/hour to work on your own open source project
- You decide what issues to work on. All the study does is randomize if you use AI or not.
- A minimum 20 hours total commitment, with totally flexible scheduling
- Your project needs to be somewhat established, and you need to be a maintainer
How it’ll work: You’ll make a list of tasks you’re planning to work on anyway (bug fixes, features, whatever you want). We’ll randomly split them into two groups - you’ll work on one group with AI assistance (using Claude 3.5) and the other group without. That’s basically it! You get to choose when you work and which specific tasks you tackle within each group.
We’re doing this because we want to measure AI’s real-world impact on development, rather than just using artificial benchmarks. Plus, you get paid to work on your project while helping advance our understanding of AI’s capabilities.
You can see more details about the experiment here.
Next steps: we’re kicking off the pilot study ASAP! If you’re interested, you can apply here.
Want to learn more about us? Check out METR or search for our name in the recent o1 model report card.