If you’ve seen DataAnnotation ads on social media and wondered whether signing up is worth your time, the answer is YES. I spent two years racking up a regular income on DataAnnotation.Tech. Here’s everything you need to know about the remote work offered by the platform.
I started working for DataAnnotation Tech during the pandemic, hoping to supplement my primary income as a freelance writer. During my time working on the platform, I was consistently pleased with the amount of work available and the pay, so I’ve become something of an evangelist for the platform, encouraging dozens of my friends and colleagues to sign up. The positive experiences they’ve reported have only reinforced my appreciation for the platform.
In the simplest terms, DataAnnotation is an AI training company. In this DataAnnotation review, I’ll break down the signup process, strategies for getting higher-paid work, my workflow as a user, and how payment works.
DataAnnotation Company Overview
I’ve seen many folks on Reddit and Facebook asking whether the site is legit or a scam, suggesting that the so-called legitimate experiences advertised on social media platforms are anything but. I can certainly understand why they might feel a bit leery since it’s admittedly difficult to find information about the company’s ownership. Due to some of the proprietary content associated with AI training, there seems to be a good deal of secrecy around these types of companies.
Although a few sites like The Verge have suggested that DataAnnotation Tech may be a subsidiary of Surge AI, finding anything concrete to support this claim has proven tricky. My confirmation of the platform’s legitimacy is based on personal experience. During my two years with the platform, I found regular work on the site, and it paid out consistently every time, on time, with no delays.
Types of Projects DataAnnotation Offers
DataAnnotation is an AI systems training platform that primarily offers data annotation and labeling projects, as well as large language model (LLM) training projects. Every project is built around machine learning; effectively, you’re working as a bot’s trainer.
In my experience, most of my jobs involved either rating bots’ responses based on specific guidelines or chatting with bots to help them create content more suited to the project goal. But I’ve also seen qualifications for math, legal, coding, and other tasks outside my wheelhouse. Image classification and other image-based tasks also occasionally appear on my dashboard.
If you perform well on qualifications and continue to take them every time they pop up, you’ll likely see a range of projects available to choose from. All of the tasks contributors work on are proprietary, so I can only speak in broad strokes when discussing my past projects. But based on my experiences, the types of work you might encounter could include casually chatting with a bot or writing out examples of perfect responses.
Although most projects require a desktop, some can be completed using a mobile phone or tablet. In rare cases, I’ve even encountered projects that require a phone. Additionally, qualifications regularly pop up looking for users with access to certain AI subscriptions or software—for example, spreadsheet or audio software, to take on a new project.
As to how long a project might last, some projects might be available only for a matter of minutes, while others could last weeks or even months.
Working on the same project for hours can become tedious, so I would often switch up my workflow. I might start with a complex, higher-paying task early in the day, when the caffeine kicks in, and shift to something more fun for a while to give my mind a break.
I also appreciated the support DataAnnotation provides employees while working on projects. Every project comes with a chatbox where users can connect with an admin or ask questions of other users, and most projects have a corresponding Slack page where users can (and, in some cases, are expected to) log in to ensure they’re on the right track.
What Kind of Schedule Is Available?
One of the best things about working for DataAnnotation Tech is the ability to customize flexible schedules. You can log in for any amount of time, no matter how short or long, and work at any time of day. As a parent of three teenagers, this makes it super easy to get in some work time despite having to take breaks to be the Mom Uber.
I’ve worked early in the morning while waiting for everyone to get ready for school, and I’ve picked up some easier tasks with a little mindless television playing in the background late at night. There’s no requirement to work for a given amount of time on any task, so users can literally just drop in, knock out some work, and sign out again.
And there is no limit on how many hours you can work. Although this isn’t true for everyone who uses the platform, I have been offered enough consistent access to projects that it would be enough to make up a full-time job, albeit contract work. During my two years with the platform, I’ve experienced a few slow freelance writing weeks and been grateful for the opportunity to pick up the slack training AI models on DataAnnotation Tech, and I have a few friends working as freelancers who say the same.
How Much Does DataAnnotation Pay?
Contributors to DataAnnotation Tech are paid either per task or hourly. In rare cases, some projects may even pay an hourly rate with per-task bonus incentives on top to help get the project knocked out quickly.
When users log in to the platform, they’ll land on a dashboard that’s sectioned into three categories: Qualifications, Projects, and Report Time. The first two categories are formatted as tables with columns for pay (generally speaking, qualifications are unpaid), number of tasks, and the date each item was created.
Qualifications allow contributors to demonstrate whether they’re a strong fit for a given project or type of project. Often, these will be annotated, “Access to higher paying projects if you pass.”
Directly under this table is the list of projects available to the individual contributor. One of the things I love most about this is the pop-up table that allows users to sort either alphabetically or in ascending or descending order of pay rate — I typically sorted my own projects list in descending order from high pay to low pay.
As a professional writer, I always performed pretty well with most language-based qualifications but would skip math or coding-based projects. Although the number of available projects can vary depending on the user, I would typically see a fairly long list ranging from around 30 to 70 projects. Going by friends’ and colleagues’ experiences on the platform, it seems to be the case that the more time you’re spending on the platform, the longer this list tends to be.
Although I was occasionally offered photo annotation tasks for about $0.02 per page, and I’ve even had a short-term LLM project pay $2 per task once, the vast majority of projects on my dashboard are hourly. On my dashboard, these projects start at $20 per hour and max out at $32. Although I have had a short-term $35-per-hour project in the past, I’ve noticed these types of projects tend to go fast. Additionally, some projects will be marked “Priority” and given an extra $1 or more per hour boost.
Reporting Pay and Cashing Out
Reporting your hours is, for the most part, on the honor system. Once you click on a project, you’ll be taken to a page with guidelines for the individual project. These will generally also include links to Google documents with additional information for each project. The platform prizes attention to detail, and it’s willing to pay contributors for the time it takes to read and thoroughly comprehend these instructions.
Occasionally, there will be a note somewhere in those instructions explaining how much time users should be spending on each item, but there’s also an admin chat at the bottom of each page for project-related questions like this for contributors who aren’t sure if they’re spending too much or too little time. Although there may be a timer on some project pages, contributors shouldn’t rely on these. Instead, it’s better to start recording your own time with an app like Toggle when you begin the work.
Sometimes, projects will run out of tasks or get paused while you’re working on them. If this doesn’t happen, you can simply close the page you’re working on when you’re done. Once you reload your project dashboard, the project you’ve been working on should appear at the bottom of the page under “Report Time.” If you accidentally input the wrong information, this is easy to fix manually. Anytime you’re working on a project page inside the platform, a running tally of the pay you have reported but not yet cashed out should appear at the top of your page if for no other reason than to inspire you to keep going.
Once you’re ready to cash this out, select the “Transfer Funds” button on the blue bar above your dashboard. On the “Transfer Funds” page, users can see their withdrawable amount, amount pending approval, and total lifetime earnings. Contributors must wait a week to cash out hourly funds or three days for per-task projects. Once your money becomes available, cashing out is as simple as clicking on the blue “Transfer Funds” button. Transfers are instant, with funds posted immediately to the contributor’s PayPal account.
How to Sign Up With DataAnnotation
I first learned about DataAnnotation while poking around the work from home corners of Reddit, and I signed up for it on a whim. The signup process includes a starter assessment, and from what I’ve heard and seen from friends and folks in the Reddit community, I would strongly advise you to take your time when signing up. Although the platform is currently looking for users in English-speaking countries, they encourage everyone to sign up since they seem to be looking to expand in the future.
When you sign up, you must enter your name, email, and phone number. You’ll also be asked to confirm that you’re over 18 years old via a third-party ID checker and sign a task confidentiality agreement.
Once you’ve completed these steps, you’ll be given the option to begin the starter assessment.
Ever since I posted a comment on Reddit confirming that DataAnnotation is a legitimate platform based on my own experiences, my inbox has been bombarded by folks who completed the assessment and never heard back despite the assessment entry point explicitly stating that if accepted, users should hear back within a week (some Redditors reported a longer wait time, so don’t give up if it takes longer). Either way, it pays to take your time and get it right.
Once DataAnnotation Tech has reviewed your work, you should start seeing paid projects appear within a week or so.
Taking the Assessments
Like most of the legitimate AI training sites that offer competitive rates, DataAnnotation hits you with some pretty heavy assessments as part of your onboarding process. Per DataAnnotation’s instructions, the starter assessment should take a minimum of 30 minutes to an hour — but take as long as you need.
Perhaps the most important thing you can do here is read through the detailed instructions several times and strive for accuracy. While working through the assessment, you’ll be asked some writing questions and given opportunities to rate sample AI responses. As the entry page clearly states, ensure your responses are given in clear, complete sentences, free of grammatical errors.
To get the best shot at future tasks:
- Read all instructions carefully
- Use clear, complete sentences
- Check for grammatical errors
You’ll be given an opportunity to take assessments based on your individual skills and knowledge. I focused on the generalized writing and logic-based assessment, but you can also opt into the assessment for coders. Once you’re through the initial onboarding assessment, DataAnnotation offers higher-level assessments on a wide range of specializations, including physics, mathematics, medicine, and law, just to name a few.
Being a language arts girly, I can’t speak to what’s on the coder tests. But I can share a bit of insight on what to expect from the generalized AI training assessment.
Data training assessments tend to touch on the types of things the tech company is looking for trainers to train into (or out of) their AI models, with most models falling under either generative (meaning they create anything from writing to art to music) or assistive (helpers like a smarter version of Alexa or Google).
To that end, the assessment might present issues like these:
- Examining internet posts to see which is more harmful or problematic based on a given criteria
- Editing text to make it better match certain criteria (for example, more friendly or neutral)
- Researching information to more accurately solve a complex question based on a given criteria
You can also expect to write a story prompt based on a very specific set of instructions. Here’s my advice for creating the best version you can:
- Pay attention to the details and be sure to work all of them into your story.
- Don’t just wing it. You’ll find it’s worth your time to Google any historical events or places that arise so you understand how to include them correctly.
- Make it creative and use imagery (appeals to the five senses).
- Use transitional language (First, then, next).
- Proofread your writing and read it out loud.
Most importantly, don’t even think about using AI to help create your content. Not only is it sketchy and dishonest, but you’re not fooling anyone, and it’s a fast track to getting kicked right out of the platform.
What Users are Saying about DataAnnotation
From what I’ve seen, many of the negative online reviews of DataAnnotation Tech express frustration with the lack of communication about why some users never heard back from the company. One Redditor wrote, “I fill everything out, picked a bunch of skills, did my background paragraph, put in my PayPal, verified my phone number, and nothing […] So I never even got the ability to pass or fail, just completely ghosted.”
Another wrote, “I don’t know how I wouldn’t have passed the assessment. I had no questions wrong at all, and I don’t want to brag, but my reading comprehension, vocabulary, fact-checking, and googling are way above average.”
While this is surely frustrating, for every comment claiming to see red flags, there are several eagerly sharing their positive, genuine experiences with the platform. As one Reddit user reported, “My husband and I both are using it as a full-time job for about 4 months now, and we haven’t had a problem and have been enjoying the flexibility.”
Enjoy It While It Lasts
Throughout my time with DataAnnotation, I was constantly coming across Reddit and Facebook posts from users complaining that the platform had axed them unceremoniously after months or even years of providing quality work. One day, they would log in, and instead of a screen full of projects, they’d see what ex-DataAnnotation folks call the Blank Screen of Death. They’d hop over to their Slack account only to realize they’d been removed without so much as an email. When they tried to contact support inside the DataAnnotation platform, the link would reroute them to the main page, telling them to check back for projects.
Throughout my two years with the platform, I brushed this off, reassuring myself that these users were probably skimming on time or had violated the rules in one way or another. I was constantly loaded up with tasks and went out of my way to document my time correctly. Most of all, I was consistently getting feedback from admins on Slack and continued to be added to new, high-paying projects. But at the same time, I kept a little Post-it note in the back of my mind not to be surprised if I ended up on the chopping block because, after all, this is Big Tech.
One week last year, I logged in to see a really high-paying project qualification, which I proceeded to ace. I was added to the Slack channel, chatted with the admin, and was looking forward to the new project getting dropped sometime that week. But instead, I logged in to find the one thing I’d always feared most: the old Blank Screen of Death.
I wasn’t shocked, and I didn’t take it personally because I knew I’d gone out of my way to perform well and make the most of my time. I was, however, disappointed to learn that several of the folks I’d encouraged to join shortly after I did were also cut around the same time (despite having no connection to me on the platform).
For a few months, I kept checking back to see if any new projects had been added, but they never were, and I soon realized I’d been removed from the Slack channel. A Facebook group for DataAnnotators also reported mass cuts around the same time, with many speculating the layoffs were due to some quarterly algorithm trawl.
I’d wondered for a brief moment if it was because I took too few hours, too many hours, or too many high-paying projects. But with as many people who were cut around the same time and with as much positive feedback as I received, I realized the most likely answer is that the platform just cycles people out every couple of years.
Why It’s Still Worthwhile
Despite getting cut, I don’t have any hard feelings toward DataAnnotation.tech as a company, even if I found the coldness of the process somewhat distasteful. When folks ask me about the platform on my still-active Reddit thread, I’m happy to recommend signing up. I just include the caveats that they should go into it realizing the job could end at any time and not rely on the gig as a permanent work solution.
DataAnnotation provided a steady income for me when I needed it, and the compensation for the type of work I was doing generally seemed fair. While some other AI platforms have been plagued with payment issues, I never experienced payment delays with DataAnnotation, and the company provided consistent work during some of the shakiest pandemic days.
Alternatives to DataAnnotation.Tech
DataAnnotation is not by any means the only good data training site out there. I’ve spoken to quite a few folks with recommendations for other LLM training sites, and I’ve been contacted by several recruiters.
Here are some of the top alternatives to DataAnnotation:
- Stellar AI
- Alignerr
- Outlier
- Mercor
- TELUS Digital
- Prolific
- X.ai
DataAnnotation Tech Review Conclusion
DataAnnotation is a great platform, even if it’s not for everyone. Although I can’t speak to the coding side, users with a strong command of Standard American English grammar and punctuation, and strong attention to detail, will be best suited for this site.
In my first year working with DataAnnotation, I made around $14,000 from that platform — a fairly decent amount of pocket change to supplement my writing income. And over the two years I worked for DataAnnotation, I racked up a whopping $37,000, even earning enough cash to buy myself a Genuine Buddy Scooter in one very lucrative weekend.
If you’re a freelance worker or full-time at a day job and you’re looking for a fairly reliable way to support your primary income, this platform may be exactly what you’re looking for.
Check out this article for more site platforms that offer short-task jobs!
