The field guide · Paid AI work

How to make money training AI as an individual

You can earn money helping AI companies evaluate answers, improve writing, check code or apply professional expertise. Platforms such as Handshake, Outlier and DataAnnotation connect individuals with this work. Start with a role that fits your skills and location, complete its screening, and measure your return across all the time you spend.

By Data Mining Masters · Published

This guide includes a Handshake referral link. We may earn compensation for qualifying referrals. Read our disclosure.

AI work connects your skills to tasks such as checking responses, then to payment for approved work.
Your contribution is the work you do: checking, explaining, writing or solving. Selection, available tasks and the project’s payment terms determine what you earn.

What are you actually getting paid to do?

You earn by helping an AI system produce more useful answers. A task might ask you to compare two responses, correct a factual error, write a challenging question, review code or explain how a professional would handle a situation. The deliverable is your judgment and the explanation behind it.

Handshake describes reviewing AI content, labeling data and suggesting improvements. Outlier lists prompt writing and response ranking. Each project supplies its own instructions, quality criteria and payment terms.

AI work, data licensing and referrals are different

AI work: you complete tasks and get paid under a project agreement. Data licensing: you authorize use of records or content you control. Referrals: you introduce someone under a separate reward program. A referral bonus is not the applicant’s wage.

Business owners looking to license an existing archive should use our business data licensing guide. The guide below is for individuals doing paid work.

Five platforms to compare

Choose by your skills, eligible location and the work you can consistently do well. These are starting points to investigate, not a ranking of earnings or acceptance odds. Public information was checked November 2, 2026. Rates and openings can change.

PlatformRoute to explorePay and the main qualification check
Handshake AIGeneral AI evaluation and subject-specific projects.Project-specific hourly rates. The current fellowship requires a supported US location and work authorization; check the role’s education and experience requirements.
OutlierPrompt writing, response evaluation and specialist feedback.Hourly or per-task work; rates vary. Its FAQ currently requires at least an associate degree, with higher qualifications for some projects.
DataAnnotationWriting, reasoning, coding and specialist tracks.Its FAQ advertises $25–$50/hour for general projects. A qualification assessment applies; advertised rates do not guarantee assignments.
MercorWork drawing on a professional specialty.Read the specific contract listing. Some applications join a talent network for future projects rather than an immediate paid opening.
AlignerrLanguage, content evaluation, coding and domain-specific work.Pay and requirements appear in job listings. Onboarding can include identity checks and job-specific assessments.

Comparison sources: Handshake, Outlier, DataAnnotation, Mercor example, Alignerr. Listed or advertised pay is not an effective hourly return across all your time.

Handshake: a starting point for students and professionals

Handshake’s AI Fellowship matches people to projects using their background and current demand. General and specialist work can be available, and the project’s rate is shown before joining. Tasks and weekly caps can change. Check the current program requirements before applying, especially your location and the qualifications for the role.

Our Handshake profile also explains its separate company-data business. The referral link at the end of this guide opens the individual-work program.

Outlier: compare the project with your actual expertise

Outlier’s FAQ describes freelance work, skill screenings and project-specific onboarding. It says rates are visible before work begins and payments are processed weekly. Read the individual listing’s location and degree requirements. An account or completed screening should not be treated as a schedule of paid hours.

DataAnnotation: choose the assessment that fits your skills

DataAnnotation offers separate general, coding, language and specialist tracks. Its FAQ lists a bachelor’s degree or equivalent experience as the generalist baseline and says the Starter Assessment can only be taken once. Prepare a quiet block of time and read the instructions carefully. Payments use PayPal; check eligibility and the actual project terms before committing time.

Mercor: distinguish an opening from a talent network

Mercor can be relevant when your professional background is the asset. Its financial analyst talent network, for example, describes AI evaluation work and explicitly says it is an application for future contracts. That example includes a profile, interview and subsequent matching. Its terms and payment arrangements should not be assumed to cover every Mercor role.

Alignerr: read the role before joining the network

Alignerr’s roles page spans writing, language and technical work. Its onboarding guide describes a resume, job application, identity verification, contract and payment setup; some jobs add an interview or assessment. Check the live listing’s requirements and payment rate. Completing signup alone does not establish your earnings.

Find a starting point

Where could your skills fit?

Choose the work you can demonstrate today. This editorial tool suggests sections to compare; it does not screen you for acceptance or show live vacancies. Your answers stay in this page and are not submitted.

Turn an existing skill into a useful work sample

A concrete example gives you something better to discuss than “I’m interested in AI.” Use an original, nonconfidential example of how you check work. The following are practice exercises we created, not actual assessment questions.

  • Writing: compare two short explanations. Identify an unsupported claim, explain why it matters and suggest a clearer sentence.
  • Coding: debug a small program you wrote. Add a test that fails before your fix and passes afterward, then explain the cause.
  • Professional expertise: create a fictional scenario in your field. Define what a good answer needs to cover and a mistake a novice might miss.
  • Language: explain how a literal translation changes meaning, tone or cultural context. Write a more appropriate version.

Keep private employer, client and customer material out of samples. Describe your skills accurately, including where your knowledge ends. When a platform prohibits outside assistance or AI tools during an assessment, complete it under those rules. A generic certificate is less useful than evidence you can perform the task being requested.

How much can you realistically make?

Your earnings depend on the work you are offered, the accepted work you finish and the payment terms. An advertised hourly rate does not tell you how many hours will be available or how much time you will spend qualifying. Use your own results to decide whether to continue.

Hypothetical example: four paid hours at $25/hour produce $100. If assessments and administration take two more hours, the return is $100 ÷ 6 = $16.67 per hour before costs and taxes. These numbers illustrate the calculation; they are not a platform offer or an expected income.

Count all your time

Effective hourly earnings calculator

Use one consistent period, such as your first week. Enter either approved earnings or money received, and keep that choice consistent. Approved money may not have reached your account yet.

Formula: (gross earnings − entered costs) ÷ (paid-work hours + unpaid hours). Defaults are hypothetical. This excludes taxes and unentered costs, and does not predict future work. Nothing is saved or sent.

Keep a simple log of application time, task time, approved earnings and payments received. Separate a quiet week with no tasks from time you actively spent waiting on support or redoing work. Compare the same period across platforms, and keep one-time setup time visible when judging the first month.

A practical first-week plan

A useful first goal is a qualified application and a clear view of the next step. You may not be assigned paid work during your first week.

  1. Pick two plausible routes. Compare the actual role, supported location and qualifications. Keep the role URL and date checked.
  2. Prepare one accurate profile. Update your resume, relevant skills and public work examples. Have required verification documents ready for the official platform’s process.
  3. Set a preparation budget. Decide how much unpaid application time you can afford. Use your practice sample to identify gaps before starting a timed assessment.
  4. Apply and record the status. Distinguish submitted, assessed, accepted, assigned, approved and paid. Avoid interpreting a talent-pool invitation as paid work.
  5. Review the offer before tasking. Confirm what counts as paid time, the rate or task fee, quality rules, payment timing and how to ask about a disputed task.
  6. Review your first results. When work arrives, track all your time and use the calculator. Continue based on the work, payments and experience you actually receive.

If no projects appear, check the official dashboard and support guidance. Spend additional time only when there is a concrete next step. Repeatedly checking an empty queue does not improve the value of the time you have already invested.

Recognize a real application and avoid task scams

Open the platform’s official website yourself and verify the role there. The FTC warns that task scams may show fake earnings, make a small initial payment, then demand a deposit to unlock more work or withdraw a balance.

  • Do not send money or cryptocurrency to unlock tasks or release earnings.
  • Do not buy an account, use another person’s identity or trust a seller promising guaranteed acceptance.
  • Use the platform’s official onboarding process for identity and payment information. A logo in a message does not verify a recruiter.
  • Keep the agreed rate, completed-work records and support messages. Report a payment problem through the official support route.

Questions before you apply

Can I do AI training work without coding?

Yes, some projects focus on writing, reviewing answers or language skills. “No coding” does not mean no qualification requirements. Compare the role and assessment with the skills you can demonstrate.

Is this passive income?

No. You are doing tasks, following instructions and producing work that may be reviewed. Treat availability and your effective hourly return as part of the decision.

Will a referral link get me accepted or increase my pay?

This guide makes no such promise. Data Mining Masters may earn compensation for qualifying Handshake referrals. Selection and your project compensation are determined by the provider.

Can I use ChatGPT to complete an assessment?

Follow the assessment’s explicit rules. Use outside tools only when allowed; do not submit prohibited assistance as your own work. Practicing a skill beforehand and receiving outside help during an assessment are different activities.

Can I apply from outside the United States?

Check each listing’s supported location and payment setup. Handshake’s current fellowship is US-based. A different platform’s remote label does not establish worldwide eligibility.

Do I need to upload personal or company data here?

No. The tools on this page run in your browser without uploads or sign-in. Applying and verifying identity happens directly with the provider. Do not upload employer or client records as work samples without appropriate permission.

Take one useful next step

See whether Handshake fits your skills.

Review the individual-work program’s current requirements and available opportunities before applying.

Explore Handshake AI work

Referral link. Data Mining Masters may earn compensation for qualifying referrals. Commercial disclosure.

Sources and approach

Published and checked November 2, 2026. We reviewed the official program pages below. Platform descriptions reflect public terms, not firsthand earnings tests. The exercises, comparison criteria and planning tools are our editorial guidance. The calculator uses the values you enter. Inclusion is not a promise of acceptance, available work or income.

Our editorial approach · Commercial disclosure