MISSION

Governance

I first wrote about AI’s risks to society in 2018, in my MBA application to U.C. Berkeley. I was worried about the pace of AI development relative to the pace of policy and what that would mean for society's ability to absorb the technology. Since then, the gap has grown wider and the stakes have gotten higher.

For ten years, I’ve studied the power of technology platforms. At Playground Global, I learned how capital shapes the formation of next-generation platforms. At Barclays, I analyzed how the largest platforms (Google, Amazon, Meta) scale, monetize, and compete. At the Optimism Foundation, I designed and implemented a governance system to reduce platform risk and increase accountability for an open-source protocol securing $14B in assets.

AI is the most consequential technology of our time, and its impact on society depends on how well we govern it. Below is my current work on AI governance, followed by the experience and experiments that inform it.

AI Governance

AI Governance is the most important thing I can imagine working on. Below is what I'm doing to get involved.

Policy

UN Global Dialogue on AI Governance — Submitted a written input advocating for institutional accountability via stakeholder voting, credible commitments, and inclusive input processes. View partial statement here.

Fellowships and Accelerators

Courses

Protocol Governance

As the Head of Governance at the Optimism Foundation, I was responsible for designing the system by which an open-source protocol securing $14B in assets gets upgraded and how the Foundation's multi-billion dollar treasury gets allocated. I joined the Optimism Foundation as the first hire and progressed to Board Member, where I continue to oversee governance work. This work has many parallels with AI governance.

Goals

• Reduce platform risk: the risk that a platform a business or user depends on changes against their interests. Frontier labs carry outsized platform risk • Promote long-term sustainability over short-term extraction in how capital is allocated. AI safety requires the prioritization of long-term considerations over short-term incentives.

Responsibilities

• Building and leading a team of ~10 and managing ~50 external contributors, scaling the Foundation's capacity • Designing the governance roadmap with technologists, political scientists, economists, and major ecosystem partners • Drafting ~100 governance proposals put to public vote. Many were controversial, and only two failed • Implementing policies in accordance with governing documents, including a Working Constitution and Operating Manual • Establishing and maintaining the governance systems' legitimacy • Running governance experiments from ideation through evaluation (see below)

Governance Experiments

At the Optimism Foundation, we pioneered an experimental approach to governance design, supported by PhD students, leading academic advisors including Andrew Hall and Robin Hanson, and external consultants.

Voter Mobilization

What we tested: What Kinds of Incentives Encourage Participation in Democracy? What we found: "Reward schemes that give people a durable stake in the community and promise a sequence of future rewards are able to broaden participation in online democracy noticeably." Read: Stanford GSB, Yale ISPS· Forthcoming paper by Eliza Oak and Andrew Hall, Political Science Research and Methods·

Deliberative Processes

What we tested: When is Deliberation Useful for Optimism Governance? What we found: "We found evidence that deliberation can be a tool for knowledge exchange and increasing social cohesion, but we found no strong evidence that deliberation shifted views toward consensus on the overall topic deliberated," Read: "When is Deliberation Useful for Optimism Governance?"

Futarchy / Prediction Markets

What we tested: Is Futarchy a More Effective Way to Make Grant Allocation Decisions than Existing Approaches? What we found: Futarchy performs "at least as well as traditional decision-making structures (i.e. a Council of experts)" at making a specific forecast related to grant allocations. Read: "Futarchy v1 Preliminary Findings." and "Experimenting with Futarchy* for Optimism Grant Allocation Decisions."

Other experiments	AI governance parallel
Protocol upgrades subject to dynamic vetos	The process of approving model updates and deployments
Independent Developer Advisory Board	Independent technical body advising on / approving model updates and deployment decisions
Metrics-driven, algorithmic capital allocation	Specification gaming
$200M+ in public-goods grant funding	Redistributing the value powerful AI creates
Bicameral and stakeholder voting	Accountability structures that incorporate non-financial interests (PBC, Trusts, etc.)
Optimistic approvals	Effectively incorporating public input
Agentic delegates	AI systems making governance decisions on behalf of humans