I've been working with Outlier AI for the past eight months, and it's been a journey worth sharing. I first came across the opportunity on Indeed and applied without expecting much. Since then, there have been both good and challenging moments. Now that Outlier AI is one of the most-reviewed platforms on GigFish, I wanted to share an honest look at what it's like to work with them. This review is especially for anyone considering a remote job in AI training.
Is it Hard to Get Into Outlier AI?
The process started with a straightforward application. I filled out the form on Outlier AI's website, uploaded my resume, and waited to hear back. A few days later, I received an email inviting me to complete a short assessment. This test was meant to evaluate how well I could follow detailed instructions and judge whether AI-generated responses matched the given prompts across several categories.
At the time, I didn't take the assessment too seriously, and a few days later I was notified that I hadn't passed. I was offered the chance to retake it, but I decided to move on and continued my job search elsewhere. Surprisingly, about a month later, I received an email inviting me to onboard with the platform as if I had passed the test. That was the first sign that Outlier AI might be a bit disorganized.
The onboarding process was very similar to the initial assessment. It included pages of project guidelines, scoring rubrics, and instructions on how to evaluate responses. After reading through the material, I had to answer multiple-choice questions and complete a few practice tasks. If I got anything wrong, the platform provided feedback so I could understand the mistakes.
Outlier AI
Outlier AI is a platform that connects domain experts with projects focused on training and improving AI models.
What is Working For Outlier AI Like?
After completing the onboarding process, which closely mirrored the real tasks, I was curious to see how the actual work would compare. The structure, format, and evaluation criteria were almost the same, so I felt well-prepared heading into my first project. Here's what the first project was like.
Starting the First Project
I received my first project assignment right after onboarding. The tasks involved evaluating AI-generated responses from a chatbot designed to help users plan their travel. For each task, I would review a user prompt followed by two AI responses. I then graded each response based on how well it followed instructions, provided accurate information, avoided unnecessary fluff, maintained writing quality, and delivered an overall helpful answer.
At the end of each task, I had to compare the two responses using a "Likert" scale from 1 to 7. A score of 1 meant Response A was significantly better, 7 meant Response B was significantly better, and 3 indicated that both were about the same. I also had to write a short paragraph explaining which response I preferred and why.
Each task was paid at $25 per hour, with a one-hour time window to complete it. About a month into the role, the pay rate increased to $30 per hour. The workflow was steady and predictable, and I had the chance to earn bonuses based on how many tasks I completed.
At that point, I was working full time on the platform and consistently earned between $1,500 and $2,000 per week. On weeks when bonuses were available, my earnings could reach $3,000 or more. Things were going smoothly, and I felt genuinely thankful for the opportunity.
Inconsistent Projects and Platform Confusion
After a few solid months of reliable work, things on the platform started to shift. I would onboard to a new project, complete the assessment, and begin working—only for the project to abruptly end a week or two later. In some cases, I was removed after receiving confusing or questionable reviews from other contractors.
At first, this wasn't a major issue. New projects tended to pop up quickly, and I was able to transition between them without much downtime. Around this time, Outlier AI rolled out a new feature called the “Marketplace,” which allowed workers to browse available projects and view details like time limits and pay rates in advance.
In theory, this was a great addition. In practice, it didn't live up to expectations. Most of the projects listed were either already full or had no tasks available. It often felt like the platform offered the illusion of choice without anything real to choose from.
I also started noticing frequent signs of internal disorganization. I received emails clearly intended for other contractors. Sometimes, I was mistakenly informed that I had been removed from a project or assigned to a new one. These errors became especially common in the summer of 2024. They happened so often that I learned to ignore them and instead logged into my dashboard to verify my actual status.
Despite the confusion, I was still earning well and rarely went more than a day or two without tasks. However, over time, the work began to decline. Projects became shorter, task deadlines were tighter, and bonuses were less frequent. I never again earned as much as I had during those peak months in late summer.
Outlier AI's Referral Program
Outlier AI has offered a referral program since the beginning. At first, users could refer friends and family by submitting their email addresses. If the referred person completed 10 hours of work, both parties were supposed to receive a bonus. In the fall of 2024, the platform expanded the program by adding referral links that made sharing easier.
The program advertises a potential payout of up to $500 per referral, which sounds impressive. In reality, it is very difficult to receive the full amount. Many users have reported seeing referral payouts listed as pending, sometimes as high as $100 or more, only to watch the amount drop to $5 without any explanation. These issues are widely discussed and have become a common complaint on the Outlier AI subreddit.
As of now, GigFish has referred over 300 people to the platform. Out of those 300, only 4 have been assigned tasks, and just 1 has received a payout. To put that in perspective, only 1.3% of referrals have received tasks, and just 0.3% have earned any payout. This trend is frequently mentioned by other users as well and reflects a broader issue with the referral system.
Outlier AI
Outlier AI is a platform that connects domain experts with projects focused on training and improving AI models.
Is Outlier AI Worth It?
Outlier AI is a real platform that pays users to help train AI models. However, it is not a stable source of income. Work availability can be inconsistent, and earning a reliable paycheck becomes more difficult over time. The platform suffers from poor organization, limited support, and a referral program that often feels misleading. While both I and members of the GigFish team have had some success with Outlier, we cannot confidently recommend it as a dependable gig in 2025.
You might be one of the lucky few who are able to earn consistently on the platform. But unless things improve, your time may be better spent exploring other platforms that pay you to train AI models. Many of these alternatives are showing signs of becoming more reliable.
That said, when tasks are available, Outlier AI remains one of the highest paying options in this space. If you have some flexibility and want to give it a try, take a look at their job board. Just keep in mind that it should be treated as a side gig rather than a full-time income stream.
