Save the World
One Saturday to build AI that makes people's lives better, and to show how you made sure it won't make them worse. Leave with a project, a team, and a room full of people who hire for this.
Come for one reason. Leave with six.
Every project shows two things.
Judges weigh both halves equally. A modest tool with an honest safety check beats a flashy demo with none.
We start with health. It is AI on hard mode.
Each edition picks one area where AI could help a lot and hurt a lot. The first is health. A wrong answer can cost a life, the language is often not English, and the nearest doctor can be a day away.
Safety grew up in medicine. The safety levels frontier labs use for their models were modeled on the biosafety levels used for dangerous pathogens. Staged release is a clinical trial by another name. "First, do no harm" is the oldest safety rule we have.
The skills are the same skills. Evals, red-teaming, calibration, interpretability and human oversight are what health AI needs most, and what the people building it are hiring for. A careful eval on a real problem says more about you than another chatbot demo.
Pick a problem worth a Saturday.
Things you could build by dinner.
Use one as written, bend it, or bring your own. Each one names its data, what you could have running by demo hour, and the part that makes it hard.
Clean data, ready in the first hour.
Public datasets, models and tools, checked by hand. Most open with no account. A few need an application, so apply early.
One Friday night. One long Saturday.
Six ways to win.
The day is the start. The strongest teams get mentors, credits and support to keep building for eight weeks, then show what they made at a demo night in the spring.
Who is putting this on.
Mox
The room. A San Francisco space for people working on hard, important problems.
The Frisconian
A newsletter that counts San Francisco. It judges the open data prize and writes up what gets built.
People who do the work
Doctors, epidemiologists and AI safety researchers as mentors and judges. Want to be one? Write to us.
Before you apply.
Save the world.
Start with one problem.
One day, a real problem, the data to work on it, and a room full of people who care how it turns out.