How a Side Hustler Makes 8500 Monthly as a Freelance AI Training Specialist

Lucas: Alright, Luna, I want to run a number by you. Eight thousand five hundred dollars a month — that's what one freelancer I've been following pulls in as an AI training specialist. Luna: Okay, that's solid. But what exactly does an AI training specialist do? Because I feel like that title could mean anything from labeling cat photos to writing philosophical essays for a language model. Lucas: You're not far off. The specific role I looked into involves writing prompt-response pairs to train large language models. Think of it like teaching a model how to answer questions, summarize documents, or follow instructions. The freelancer works through platforms like Scale AI and Surge AI. Luna: So it's not just clicking bounding boxes on images. It's more like being a tutor for an AI. Lucas: Exactly. The pay reflects that. This person makes roughly $40 to $60 per hour, depending on the complexity of the task. They work about 30 to 35 hours a week, which puts them around $8,500 a month. But it's all project-based — there's no guaranteed minimum. Luna: That's a pretty decent hourly rate for freelance work. What kind of background do you need to land those gigs? Lucas: Surprisingly, you don't need a computer science degree. The key skills are strong writing, attention to detail, and the ability to follow very specific guidelines. In fact, a lot of these platforms look for people with backgrounds in journalism, editing, or even teaching — because you're essentially evaluating and improving text. Luna: So a liberal arts degree might actually be an asset here. That's refreshing. Lucas: Right. The person I spoke with has a degree in English literature. They started doing freelance copywriting, then pivoted into AI training when they saw the rates were higher and the work was more consistent. Now it's their full-time side hustle. Luna: How do you actually find these gigs? Is it just applying on Scale AI's website, or is there a more strategic way in? Lucas: Most people start by applying directly on platforms like Scale, Surge, or Appen. They have qualification tests that assess your writing and reasoning. Once you pass, you get added to a pool of contributors. But the real money comes from specializing — for example, focusing on coding prompts or medical domain tasks, which pay more. Luna: Specializing bumps the rate. That makes sense. So what does a typical task actually look like? Give me a concrete example. Lucas: Sure. Let's say the project is about improving a model's ability to summarize financial news. You might be given an article and asked to write three different summaries at varying lengths — one sentence, one paragraph, and a bulleted list. Then you also rate the model's existing summaries on accuracy and clarity. Luna: So you're both creating examples and evaluating the model's output. That's almost like a quality assurance role. Lucas: Exactly. And the guidelines can be incredibly detailed. Some projects have style guides that are 50 pages long. You have to be meticulous. That's why people with editing backgrounds do well — they're used to following style rules. Luna: Fifty pages? That's intense. How long does it take to go from applying to actually getting consistent work? Lucas: For this freelancer, it took about three weeks to pass the initial tests, and then another month before they were assigned to a steady project. During that time, they were only getting a few hours of work per week. But once they proved themselves — high accuracy, good feedback — the hours ramped up. Luna: So there's a ramp-up period. That's important for listeners to know — you don't just start at $8,500 a month. Lucas: Right. In the first two months, they made around $2,000 total. It wasn't until month four that they hit $5,000, and by month six they were consistently above $7,000. The $8,500 month came after about eight months of building a reputation. Luna: That's a realistic timeline. I think a lot of people hear big numbers and assume it's immediate. So what's the ceiling? Can you scale beyond $8,500? Lucas: Some people do. If you get onto a high-paying project — like coding or advanced math — you can earn up to $80 an hour. But the trade-off is that those projects are less common. The more typical ceiling is around $10,000 to $12,000 a month if you're working full-time hours and have a few years of experience. Luna: And what about the long-term outlook? With AI models getting better, won't the need for human trainers eventually shrink? Lucas: That's the big question. Right now, the demand is still high because models need continuous fine-tuning and new domain-specific data. But it's likely that the market will become more competitive, and the rates may drop. The freelancer I talked to is already planning to pivot into AI consulting — helping companies set up their own training pipelines — which pays more and is more sustainable. Luna: So it's a side hustle with a shelf life, but it can be a stepping stone to something bigger. Lucas: Exactly. And honestly, that's a pattern we see in a lot of side hustles. You enter at a high-paying entry point, build skills and a portfolio, and then move into a higher-value service. Luna: Speaking of building something bigger, I want to circle back to something you said earlier about the value of this show. We've now done 70 episodes, and I think one reason we keep going is that listeners tell us these conversations actually move their work forward. And we've kept the show completely ad-free from day one. Lucas: Yeah, that was a deliberate choice. No ads, no sponsors, no interruptions. If you've found value in episodes like this one — maybe you're considering starting your own AI training side hustle — and you want to support that choice, there's a simple way to do it. It's buy me a coffee dot com slash fexingo. That's all lower case, no spaces. Luna: And that goes directly to keeping the show independent. No pressure, but if it's made a difference for you, it helps us keep doing what we do. Lucas: Alright, back to the AI training specialist. So let's talk about the practical steps if someone wants to start tomorrow. Luna: Yes, let's get into the nuts and bolts. What's the first thing you do? Lucas: Go to Scale AI's website and look for their 'Remotasks' platform. That's the entry point. Sign up, take the qualification test. It's a mix of multiple choice and free-response questions. Do your best, but don't stress if you don't pass on the first try — you can retake after a waiting period. Luna: So it's not a one-shot deal. That's good. And while you're waiting, what should you do to prepare? Lucas: Read their style guides and example tasks that are often available in community forums. Also, practice writing concise, accurate responses to prompts. You can even use ChatGPT to generate sample prompts and then critique its output — that's essentially the skill you need. Luna: So the tool you're training is also the tool you can use to practice. That's kind of meta. Lucas: It is. And one more thing — if you can, specialize in a domain you already know well. If you're a nurse, focus on medical AI training. If you're a lawyer, legal prompts. That domain expertise commands higher rates and makes you harder to replace. Luna: Let's talk about the downsides. What are the main challenges? Lucas: The biggest one is inconsistency. You might have weeks with 40 hours of work, then a week with only 10. Projects end suddenly. Also, the work can be mentally exhausting — it's very focused, repetitive, and requires constant decision-making. Luna: And there's no benefits, no paid time off. So you have to manage your own finances and schedule. Lucas: Right. That's true for any side hustle, but especially for AI training because the work is so deadline-driven. The freelancer I spoke to sets aside 30% of every paycheck for taxes and keeps a three-month emergency fund. Luna: Smart. So if someone is listening and thinking, 'This sounds like me,' what's the one thing you'd tell them to do right now? Lucas: Go take that qualification test today. Don't wait until you feel ready. The test itself will show you what you need to learn. And even if you don't pass, you'll have a clearer picture of the skills required. That's the first step. Luna: And for those who are already doing this kind of work, is there a next level? Lucas: Yes. Look into becoming a project manager or reviewer on these platforms. Reviewers get paid a premium to check the work of other trainers. Or start offering AI training consulting to small businesses that want to build custom models — that's where the real money is. Luna: So the side hustle can be a launchpad, not just a destination. Lucas: Exactly. And that's the mindset — treat it as a stepping stone. Build the skills, save the money, and then pivot into something that gives you more control and higher income. Luna: Alright, I think we've given listeners a solid roadmap. If you try it, let us know how it goes.