Our purpose
Science belongs
in the open.
The Open Research is a place to understand AI research together: to discover important work, question its claims, and share the evidence.
A paper is the beginning
of a conversation.
Research is more useful when people can see how it works. The approach, the data, the compute, the code, the failures—all of it belongs in the conversation.
We’re building a curated starting point for newcomers and a shared reading room for researchers and practitioners. A place where thoughtful criticism is welcome, and a paper’s reputation can be examined alongside its evidence.
Ask better questions.
Our proposed review frameworkA single score can hide a lot. These five dimensions keep different judgments visible. Each review should explain its reasoning, cite evidence, and disclose relevant conflicts of interest.
Scientific rigor
Do the methods, baselines, and experiments support the claims?
Reproducibility
Can an independent team reproduce the reported results?
Implementation
Is the code correct, documented, and practical to run?
Transparency
Are the data, compute, limitations, and evaluation details disclosed?
Usefulness
Does the work help us understand a problem or solve one in practice?
Critical. Constructive.
Open to revision.
Critique the claim, not the person. A failed reproduction is useful evidence when its conditions are documented. A missing review is not a negative review. Strong conclusions need enough independent evidence.
Scores, review counts, and uncertainty should be shown together. Authors should have room to respond, and conclusions should change when better evidence arrives.
Find a paper to explore