How to apply for Coordination and Accountability

About Foresight Institute

Foresight Institute is a San Francisco-based research nonprofit focused on early-stage, interdisciplinary science and technology that legacy institutions overlook. It is a hub for long-term thinking on transformative technologies like AI, nanotechnology, and biotechnology. Working here means shaping the future of safe, decentralized coordination in AI—a rare chance to influence a field at its inception.

About the role

This role is about building the technical and social infrastructure for safe, decentralized coordination among multiple AI agents and humans. You will design protocols, incentive mechanisms, and oversight systems that resist gaming and manipulation without a central authority. Your work will directly contribute to the emerging field of multi-agent alignment and decentralized governance.

A typical day

A typical day might involve brainstorming coordination protocols with a small interdisciplinary team, coding a prototype for credit assignment, and reviewing recent papers on multi-agent alignment. You'll also spend time writing research proposals or engaging with the broader Foresight community to refine ideas. Expect a mix of deep technical work and collaborative strategy sessions.

Who Foresight Institute is looking for

• Deep expertise in multi-agent systems, mechanism design, distributed systems, or AI safety, with a track record of research or implementation.
• Experience designing coordination protocols or incentive mechanisms that are robust to adversarial behavior and gaming.
• Familiarity with decentralized governance, Schelling points, or boundary-based coordination concepts.
• Ability to work independently in a remote, research-driven environment and communicate complex ideas clearly to interdisciplinary teams.

Tips for this application

• Highlight any prior work on cooperation, decentralized governance, or multi-agent coordination—even if it's theoretical or from adjacent fields like blockchain or game theory.
• Show you understand Foresight's focus on 'too early-stage or interdisciplinary' tech by referencing specific Foresight projects or publications in your application.
• Demonstrate concrete experience with mechanism design or distributed oversight by describing a project where you built or analyzed such a system.
• Emphasize your ability to work remotely and asynchronously, as Foresight is a distributed team with a research-first culture.
• Use the cover letter to propose a specific research direction or prototype you'd like to pursue within the role, aligning with Foresight's mission.

What to cover in your cover letter

['Your specific technical contributions to multi-agent coordination or decentralized alignment, with examples.', 'Why decentralized, non-centralized oversight is critical for safe human-AI collaboration and how your background addresses it.', 'How you would approach designing incentive mechanisms that resist gaming and manipulation in open platforms.', "Your alignment with Foresight's long-term, interdisciplinary approach and your desire to work on early-stage, high-impact research."]

Research before applying

• Read Foresight Institute's research agenda and recent publications on AI safety, coordination, and decentralized technologies.
• Explore their 'AI Alignment' and 'Neurotech' programs to understand how this role fits into their broader mission.
• Look into the work of Foresight's researchers and affiliates, such as those involved in multi-agent systems or mechanism design.
• Review the concept of 'boundary-based Schelling points' and how it relates to Foresight's approach to coordination.

Likely interview topics

Based on the job description, expect questions about:

• How would you design a credit-assignment protocol for a collective of AI agents and humans that prevents free-riding?
• What are the key challenges in distributed oversight of AI agents, and how would you address them without a central authority?
• Can you discuss a time you applied mechanism design to a real or theoretical coordination problem?
• How do you envision boundary-based Schelling points enabling multi-agent coordination and alignment?
• What metrics would you use to evaluate the safety and robustness of a decentralized coordination system?

Common mistakes to avoid

• Focusing only on centralized AI safety solutions—Foresight explicitly seeks decentralized, non-centralized approaches.
• Neglecting to mention mechanism design or incentive alignment, which are core to the role's requirements.
• Submitting a generic cover letter that doesn't reference Foresight's unique focus on early-stage, interdisciplinary research.