Apply for Research Support

AE Studio's survey found that only about 10% of AI researchers believe current approaches will solve alignment in time. Most alignment funding goes to mechanistic interpretability, evals, and AI control. All three are necessary, and practitioners in each community acknowledge that none of them will solve alignment alone at the scale of recursive self-improvement.

Recursively self-improving systems are approaching. The alignment properties that matter most are the ones that persist and strengthen as systems improve themselves. That requires high-risk, high-reward R&D on new directions, and the research cycle is long enough that the work needs to start now.

We fund alignment researchers who have ideas they can't pursue within their current institutional funding, and we bring in technical experts from other fields whose perspectives could produce breakthroughs. We accelerate their work with dedicated engineering teams that move ideas from concept to published results.

You're already in the space with ideas you can't pursue. Many alignment researchers have ideas just as impactful as their current work, but their organization or funding only covers a narrow slice. We fund those ideas and provide engineering teams that can execute on them with as little as one hour of your time per week.

You work outside the alignment field, but you see a connection others are missing. Many of the smartest people who could contribute to alignment haven't worked on it yet. Cognitive neuroscience, mathematics, physics, and other fields hold perspectives that could reshape how we approach the problem. If you have an idea along those lines, we want to hear it.

The idea and your ability to execute. What matters most is technical substance and a clear theory of change. We've funded researchers from top institutions and from outside the field; the common thread is that their ideas hold up under scrutiny.

Field-making potential. We look for work that opens entirely new directions. Ideas that generate more ideas, not incremental improvements on well-studied problems.

Interdisciplinary leverage. Breakthroughs often come from unexpected connections across domains. If your background gives you a lens others in alignment don't have, we want to hear from you.