Microtasks and AI training jobs: where they pay and how much

What microtasks and AI training are

Artificial intelligence models need people: to judge whether an answer is right, compare two options, record a voice, check a translation or code. The tasks are small, handed out to many people at once, and paid per task or per hour.

Platforms differ in their requirements. Some, like Clickworker, give simple tasks to a wide range of people. Others, like DataAnnotation or Outlier, recruit by assessment and qualifications and pay more. Prolific combines both: regular studies for everyone and separate, better-paid AI work for those who pass an assessment.

Clickworker: the easiest entry

Clickworker speaks of more than a million microtasks and says it is one of the providers for the UHRS system with a large number of tasks. Tasks include AI training, surveys, voice recordings, photo contests and short phone videos. You see the pay before you start, and an accepted job is reserved for you for a fixed time, so you can do it offline and submit it later. The platform works on any of your devices, and there is an app for phones.

Payouts are weekly, Wednesday to Friday: PayPal from 10 euros, Payoneer from 20 euros or dollars, SEPA for EU countries from 10 euros. But Clickworker temporarily switches off registration for some countries, and before you start it asks you to check that at least one payout method is available to you.

Prolific: AI taskers and domain experts

Prolific has a separate group of participants called AI taskers. The platform picks people with high approval rates and strong past performance and invites them to a two-part assessment that simulates real AI tasks. Those who pass see extra, better-paid studies evaluating and training AI models in their dashboard. Failing does not affect your account.

There are also domain experts: Prolific invites people with the right education or profession to verification, which may be a skills assessment, a document check such as a medical licence, or a review of qualifications. You do not apply separately: external adverts are for people new to Prolific, and a second account is not allowed.

DataAnnotation: hourly rates

DataAnnotation publishes its rates. General projects start at $25–30 an hour: rating chatbot responses, comparing AI outputs, testing image generation; they need strong writing and critical thinking. Multilingual projects start at $20 an hour: translation, localisation and cross-language annotation. Coding, sciences, law, finance and medicine start at $50 to $100 or more an hour.

The baseline requirement is a bachelor’s degree or equivalent real-world experience. Each track has its own starter assessment, and the platform says plainly that not everyone gets in. Payouts go to PayPal within a few days of a request, so make sure you give the correct PayPal email address. Hours are flexible, with no minimum.

What you need to pass the assessment

DataAnnotation describes the requirements for each track. For general projects the key is strong writing and critical thinking: you have to explain why one chatbot answer is better than another. For coding, experience with Python, JavaScript, HTML, C++, C# or SQL. For sciences, a master’s or PhD in mathematics, physics, biology or chemistry, or a bachelor’s degree plus more than 10 years of professional experience. For law, finance and medicine, the relevant licence.

The platform stresses that it values on-the-job expertise as much as formal education. You choose the starter assessment that matches your expertise, so do not pick a track where you have no real knowledge.

Outlier: the Scale AI platform

Outlier is a platform operated by Scale AI that connects experts with AI companies to give feedback on language models. It mentions 50 countries. Onboarding consists of account verification, qualifying for a project through assessment tasks, and then the work itself.

To apply you need a valid ID and a mobile number from your country of residence, a current resume and a LinkedIn profile showing your education and experience. Rates depend on expertise, project complexity and location, and you see the rate before starting. Payouts are weekly on Tuesdays for the previous Tuesday to Monday: PayPal, Airtm or ACH bank transfer.

Is it available from Ukraine

Prolific is not available from Ukraine, because Ukraine is not on its country list. Clickworker may temporarily close registration for some countries, so check the sign-up form. DataAnnotation and Outlier do not publish country lists on their public pages, so we cannot confirm availability for Ukraine.

Outlier requires an ID and phone number from your country of residence, and for people on a visa, a visa that permits work. For Ukrainians abroad that means signing up with documents from the country where you live.

The pay-to-earn scheme

The popularity of "AI training" has fuelled fake task jobs. The US Federal Trade Commission describes them this way: a messenger text about online work with few details, then tasks such as "app optimisation" or "product boosting", liking videos or rating product images. Tasks come in sets of 40, small payouts may arrive at first, and then you are asked to put in your own money to complete the next set.

According to the FTC, there were no reports of these schemes in 2020, 5,000 in 2023 and about 20,000 in the first half of 2024 alone. Payment is usually demanded in cryptocurrency.

• genuine platforms do not charge sign-up or training fees; even DataAnnotation says so
• real work does not arrive as a first message on WhatsApp or Telegram from a stranger
• any request to top up a balance before you can be paid is a scam

How to choose your format

With no experience, start with Clickworker microtasks if registration is open in your country. If you write well in English and think critically, try DataAnnotation’s starter assessment for general projects. If you are a programmer, scientist, lawyer or doctor, the highest rates are in the specialised tracks. And if you already have a Prolific account in a listed country, watch for an invitation to the AI tasker assessment.

Remember the rules too: using ChatGPT or other AI assistants for answers in Prolific studies is forbidden unless the researcher explicitly asks. User Interviews advises against answering screeners with ChatGPT, and CleverX checks answers for AI use and may hold payment because of it.