For teachers

The training data for good pedagogy is thin. Most of what a language model sees during training is content written for a reader who already understands the material. Very little of it is content written for a learner encountering a concept for the first time, with the specific misconceptions a real classroom throws up, at the right cognitive level for the age group.

That gap is exactly why frontier AI labs are quietly hiring practicing teachers as expert contributors. The people who have spent years watching a fourteen-year-old confuse voltage with current, or a ten-year-old skip a step in long division, are the people who can teach a model to catch the same errors and respond well.

What the work looks like

• Response evaluation. You review model outputs on educational prompts (explaining a concept, marking student work, correcting misconceptions) and score them against the criteria a working teacher would apply.
• Question authoring. You write worked examples, lesson plans, and diagnostic questions at a specific age level and difficulty. Depth matters more than volume.
• Misconception surfacing. You document the specific errors students in your subject make, and the moves a good teacher uses to correct them.
• Rubric review. You audit the criteria being used to evaluate educational outputs and flag anything that would let bad pedagogy through.

The work is asynchronous. You set the pace. Reporting is structured but not oppressive.

What labs actually look for

Recent classroom experience in a specific subject at a specific level. Teacher training or licensure where applicable. Written English at a professional standard. Willingness to explain your reasoning, in writing, as if you were showing another teacher.

Subject and level matter more than tenure. A mid-career maths teacher in Karachi with a strong record is often a stronger candidate than a former head of department who has been out of the classroom for years.

Where the pay lands

Rates are in USD and are set to Western scales. For teachers in emerging markets, the difference against local salary is often significant. For teachers in higher-cost markets, the rate is competitive with private tutoring at the specialist end.

How to enter

Sign up, upload a CV that shows your subject, level, and recent teaching, and let the platform surface the currently open roles in educational annotation, evaluation, and content authoring. Apply where you can genuinely deliver. Skip where you cannot.