$100–150/hr · Mercor · Part time, 40 hours a week
An experienced attorney who trains AI models by reviewing legal outputs for accuracy, reasoning, and compliance with procedural requirements.
What you would do
- Draft reference legal documents from detailed scenarios that serve as AI evaluation benchmarks
- Review AI-generated contracts and discovery materials to verify provisions accomplish intended purposes
- Evaluate whether legal reasoning reaches correct conclusions given specific jurisdictions or forums
- Check whether cited authorities actually support propositions they are cited to support
Who they want
- J.D. or LL.B. with active bar admission
- Real litigation, transactional, or regulatory experience where you personally conducted analysis
- Ability to explain your legal reasoning in clear, detailed writing
- Comfort flagging ambiguous instructions and working with open-ended analysis tasks
What the interview asks about
1.Procedural vs substantive legal error
AI models often state black-letter law accurately while missing jurisdictional nuances that determine actual case outcomes, and interviewers need to know you catch this.
For example: “An AI motion stated the summary judgment standard correctly but didn't account for your jurisdiction's different standard of review. How did you explain why it would fail despite correct citation?”
2.Privilege scope and application
Misapplying privilege in discovery carries professional responsibility consequences and compromises the entire discovery process.
For example: “An AI draft claimed attorney-client privilege for emails with five business managers copied. What was the problem and how would you teach the system the correct privilege scope?”
3.Case citation accuracy
A case that appears favorable might actually hold something narrower or contrary when read carefully, and models need training on citation verification.
For example: “An AI cited a published case for the proposition that non-compete agreements are enforceable, but the holding actually said they're unenforceable under state statute. How did you explain that error?”
4.Reference document completeness
The reference materials you create determine what the AI learns about complete, correct legal work, so accuracy matters enormously.
For example: “Given a fact pattern with multiple parties and overlapping contract claims, how did you decide which causes of action to include versus exclude as immaterial?”
5.Explaining technical judgment
Non-lawyer researchers need to understand why your professional judgment reached specific conclusions about model outputs based on practice experience.
For example: “You flagged an AI's indemnification language as technically sound but commercially impractical in transactional negotiation. How would you explain that distinction to people without legal training?”
A task you may get
Review an AI-drafted contract provision, identify whether it accomplishes its stated goal, flag any privilege issues, and explain your reasoning in writing.
How to prepare
- Walk through a recent complex matter from your practice: the facts, key legal issues, your analytical process, and where jurisdictional factors affected your conclusion
- Prepare an example of legally correct language that would still fail given procedural or forum-specific context
- Refresh your knowledge of privilege rules and discovery standards in your practice area
- Think through how you would explain legal concepts to an audience unfamiliar with law
The facts
- Pay
- $100–150/hr
- Hours
- Part time, 40 hours a week
- Where
- Remote
- Field
- Law
- Role type
- Talent network
- Posted
- 9/15/2026
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.