Mathematics

Mathematics — Up to $40/hr

Tier: STEM

Mathematics subcontractors review the calculus, algebra, statistics, and proofs that AI models produce. Models often get the final answer right while the working is fabricated — and equally often, the presentation looks airtight while the answer is off. Both matter, and we test for the disposition that catches both.

Key facts

• Practice test: free, 15 questions (12 domain + 3 behavioural). Available now.
• Cognition Test: 60 questions (40 domain + 20 behavioural). Fee published at checkout.
• Pass mark: 78% — no time limit.
• Rate ceiling: Up to $40/hr
• Tier: STEM
• Retake policy: Yes — no lockout after a failed attempt.

About the specialty

Mathematics review is harder than it looks because models often get the answer right while the working is fabricated. A correct answer with bad working is still a failed output — because it can't be trusted at scale. Your job is to find where the reasoning broke, not just whether the final number matches.

Topics covered in the test

The Cognition Test for Mathematics draws from a 150-question bank covering algebra, calculus, statistics, probability, discrete mathematics, and proof verification.

Difficulty mix

10 easy / 20 medium / 10 hard in the domain section. Behavioural section unweighted by difficulty.

Sample question topics

Verifying a model-generated proof step-by-step; spotting algebraic slips three steps back; catching a statistics answer that uses the wrong distribution; identifying a calculus solution where the differentiation is right but the integration is wrong.

What we test for

Working-checking over answer-checking. Anyone with a calculator can verify a final number. We test for the disposition that traces the reasoning and finds the break.

After you pass

Qualified Mathematics subcontractors are matched to live AI review work within seven days. Rate ceiling is up to $40/hr at the STEM tier.