For researchers

Frontier AI labs have quietly become one of the most active buyers of active research expertise across the sciences and social sciences. The reason is straightforward. Models are being pushed into domains where the training data is thin, out of date, or written for a general audience rather than a working researcher. Filling that gap requires researchers currently active in the field: reading recent literature, running experiments, teaching graduate students, or advising on policy.

This includes classical academic disciplines (physics, chemistry, biology, mathematics, statistics, computer science, economics, political science, sociology, history, philosophy, linguistics) and applied research fields (public health, epidemiology, climate science, materials science, energy systems, security research). If your title has "researcher," "scientist," or "postdoc" in it, or you hold a graduate degree and are actively engaged in the field, you are in scope.

What the work looks like

• Response evaluation. You review model outputs on prompts inside your specific research area and score them against what a competent researcher would say.
• Question authoring. You write research-relevant prompts at the specificity of a graduate-level exam or a peer-review comment.
• Literature grounding. You verify specific claims against primary literature and flag when a model has generated a plausible-sounding citation that does not exist.
• Adversarial testing. You construct prompts where the model is likely to blend related but distinct sub-fields, confuse similar theoretical frameworks, or overstate a result.
• Rubric review. You audit the criteria being used to grade outputs in your field and flag anything that would let common misconceptions through.

The work is asynchronous. Two to four hour blocks. Structured deliverables.

What labs actually look for

Graduate training or verifiable active engagement in your field. Recent publications, current teaching, or current advisory work all count. Written English at the level a research paper expects. Region, institution, and passport do not matter.

They do not look for AI experience. Almost no researcher hired for this work has ever trained a model.

Where the pay lands

Rates are in USD and are indexed against Western research salaries. For researchers in emerging markets, this often exceeds a full postdoc stipend on a per-hour basis. For researchers in higher-cost markets, it is competitive with consulting.

How to enter

Sign up, upload a recent CV that makes your field, degree, and active work visible, and let the platform surface open roles that match your specific expertise. Apply where the fit is honest.