About
Pragmatic Frontier is a Super Intelligence organization that applies AI/SI research to near-future products.
Our work spans decision models, personal systems, and new ways to measure intelligence. We are interested in models that know when to decide and when to defer to systems that learn in the open, and interfaces that make uncertainty useful.
Our aim is to put frontier capabilities to work on problems that require judgment, which means moving between research, prototypes, and the small details tied to everyday use.
Role
Work across research and product to find out which AI research holds up when a product depends on it.
Start from a question, build a way to measure it, and follow the useful results through to a model or a benchmark. Work may move between a decision model, an evaluation, and the data connecting them. You’ll need to test what happens when a model is confident and wrong, or when experts disagree, and find the limits of a result before someone builds on it.
We want researchers who can think through the evidence and care about the decision it informs.
Process
Share a short introduction, links to relevant work, and a research question you want to explore.
Show us how you think. Relevant work could be a paper you wrote, a benchmark you built, or a result that changed your mind. We are interested in the reasoning behind the finding, more than we’re interested in the finding.