Quick answer
AI training jobs pay people to make AI models better: writing model answers, rating AI replies and checking work in their own field. On 7 November 2026 the median listed rate was $65 an hour across 1,054 open roles, with specialists in law, medicine and finance listed highest. Most roles are remote contracts with no guaranteed hours. You can browse live roles from 11 platforms on AI Global Work, our sister site.
AI companies need people. Every model you use was shaped by humans who wrote example answers, compared responses, spotted mistakes and explained what a good answer looks like. That work is now a real, paid category of remote job, and in some fields it pays very well.
It is also a category with real catches, and plenty of people selling a fantasy version of it. This guide covers both.
The pay figures here come from AI Global Work's live pay page, which collects every role from each platform's public listings daily and uses the hourly rate each platform lists. We checked them on 7 November 2026. They change every day, so treat them as a snapshot. AI Global Work is our sister site: it is free to use and earns a referral fee when you join Mercor or micro1 through its Apply links. It says pay figures and ordering are never affected by that.
What AI training work actually involves
"AI training" covers a few different kinds of work, and the job titles vary by platform.
- AI tutor or trainer. You write the kind of answer a model should give, often in your field, so the model can learn from it.
- Rater or evaluator. You compare two or more AI replies, rank them, and explain why one is better. Accuracy, safety and tone are the usual criteria.
- Expert reviewer. You check AI answers in a specialist area, such as a legal question, a medical case or a piece of code, and correct what is wrong.
- Data collection. You supply material a model learns from: photos, recordings, videos or written examples, usually against a detailed brief.
The common thread is judgement. The work rewards people who read instructions carefully, notice small errors and can explain their reasoning in writing.
What it pays
On 7 November 2026, AI Global Work's live data showed:
- A median listed rate of $65 an hour across 1,054 open roles that list an hourly rate.
- 383 roles listing $100 an hour or more.
- 1,243 open roles in total across 11 platforms.
The field you work in matters far more than the platform. Median listed rates by field that day:
| Field | Median listed rate |
|---|---|
| Law | $105/hr |
| Medicine | $98/hr |
| Finance | $90/hr |
| Creative and media | $70/hr |
| Coding | $65/hr |
| Generalist | $60/hr |
| Writing | $53/hr |
| Maths and science | $50/hr |
| Languages | $45/hr |
| Data collection | $26/hr |
If you have professional expertise, this is where AI training work gets interesting: a lawyer, doctor, accountant or experienced developer can find listed rates well above the median. If you do not, generalist, writing, language and data collection roles are the realistic starting points, and the pay reflects that.
The catches nobody puts in the advert
Listed pay is not take-home pay. AI Global Work's own pay guide is direct about this, and so are we:
- Most roles are contracts. You cover your own taxes and there are no benefits.
- Hours are not guaranteed. Projects start and stop, and some weeks have no tasks at all.
- Some projects pay per accepted task, so work a reviewer rejects can go unpaid.
- Onboarding, interviews and assessments are usually unpaid.
So think of AI training work as a well-paid side income or a flexible bridge between jobs, not as a guaranteed salary. People who do well tend to keep it alongside other work until a project proves steady.
How to get started, step by step
- Pick the field where you are strongest. Your expertise is the thing being paid for. A nurse should look at medicine roles before generalist ones; a bilingual speaker should look at language roles.
- Browse what is open right now. AI Global Work's role list lets you filter by field and language, and shows the pay each platform lists. If you are not sure where you fit, its Find my fit quiz asks five questions and suggests roles.
- Read the eligibility before you apply. Many roles are limited to certain countries, require a degree or professional licence, or need a particular language. Eligibility varies by role.
- Apply on the platform itself. You sign up with the platform that runs the project, not with a middleman, and expect an assessment or interview. Some platforms interview with an AI interviewer.
- Treat the first project as a trial. Read the guidelines twice, work slowly at first and keep your accuracy high. Quality scores usually decide whether you are offered more work.
How to spot a scam
Because the work is remote and well paid, it attracts scams. One rule covers almost all of them: money only ever flows towards you. A real platform never asks you to pay to get work, buy a starter kit, pay for "certification" before your first task, or move money for them.
Other warning signs: offers that arrive unsolicited on WhatsApp or Telegram, pay that is far above every listed rate for easy work, pressure to decide today, and companies you cannot find anywhere independent. Our free course lesson on spotting annotation job scams covers the five tests to apply to any platform.
If you are using AI Global Work, check that you are on aiglobalwork.com and that you apply through the platform's own site.
Which platforms hire
AI Global Work tracks 11 platforms: Mercor, micro1, Alignerr, xAI, Scale AI, Terac, G2i, Handshake AI, Anyone AI, 10x Team and HumanSignal. They differ a lot in how they screen people, how they pay and which fields they focus on. AI Global Work publishes plain-English guides on several of them, for example whether Mercor is legit and Mercor vs micro1, including the downsides of each.
We are not affiliated with any of these platforms. Their pay, eligibility and terms are set by them and change often, so always check the role page before you apply.
Where LearnWithZavi fits
LearnWithZavi is a free AI school, and a few of our courses are built for exactly this kind of work:
- AI Annotation Jobs explains what the work involves, what it pays, how to judge a platform and how to move up to better-paid specialist work, with an optional practice lab of realistic exercises.
- Build an AI Portfolio Employers Can Verify helps you show your judgement with real examples, which helps with expert roles.
- AI-Powered Job Search covers CVs, cover letters and interviews if you are also looking for a permanent role.
All three are free, need no signup, and end with a free certificate of completion. For the wider picture of which AI skills employers want, read AI skills for jobs.