A human purpose for more capable machines.

Follow the benefits all the way to people.

We care about what people can do, the choices they control, and the benefits they receive. A faster task is a starting point. Time to learn, stronger security, meaningful work, and freedom to care for others belong in the account of progress too.

From authored skills to human choice.

The work began as Skill Asset Protocol, exploring reusable expertise, attribution, and compensation. Wallet Kernel followed with a concrete engineering question about authority over AI spending.

Human Choice gives those questions a shared purpose. The initial research focus is people whose expertise and everyday work are being reshaped by AI. The earlier work remains available with its historical context and evidence limits.

Build, inspect, learn, revise.

We connect each system to a human question. We distinguish a principle from a hypothesis, an experiment from a finding, and a payment record from evidence of useful work.

We intend to publish methods and results, including negative findings and corrections. The first human-outcomes studies are proposed; current engineering evidence is described on the Systems page.

Bring a real question.

Researchers, builders, and people using AI in their work can help sharpen the questions. A useful starting point is a specific situation: what changed, who benefited, what became harder, and what would have helped.

Research discussion and implementation work currently take place in the public GitHub repository.