Why Forall Labs exists

Forall Labs is an open research community in AI. It exists because the gap between who can do research and who is allowed to has become too wide to ignore.

The problem

If you want to do serious AI research, the usual route runs through an institution: a university group or a company lab, somewhere with compute and a name. That route is narrow, and it is getting narrower. Plenty of people who could contribute are outside it. They are between degrees, in industry jobs that are not research, in countries without a well funded department nearby, or simply without the affiliation that gets a submission read.

Meanwhile, a growing share of the field's most consequential work happens inside organisations that are not obliged to show anyone how it was done. That is a poor arrangement for a technology this general.

What we are building

An open research community that holds itself to a short list of principles: publish everything, choose work by who it helps, decide together in the open, credit everyone who contributes, and report what did not work. They are on the front page because they are meant to be held against us.

In practice, anyone can join, anyone can propose a project, projects are staffed by whoever shows up, and everything gets published, including the work that did not pan out.

What we want to produce

The goal is not a long list of papers. It is a working community and projects that finish, with code and data that someone else can run. A lab that publishes three reproducible things is more useful than one that announces thirty.

We would rather be judged on that than on anything we say about ourselves.

How to take part

If this sounds like something you want to exist, the most useful thing you can do is join or propose a project.

∀ is the mathematical symbol for “for all”. It is the whole argument in one character.