Idea #17: AI Training Work - $30-$85/Hour Rating AI Responses

Hey buddy,

Today's WiFi Moolah idea is for anyone with expertise in literally anything - finance, medicine, law, writing, even basic data entry.

Remote AI Training Jobs (Mercor & Similar Platforms)

The Idea: Sign up on platforms like Mercor, complete AI interview, get matched with tasks like rating AI responses, grading model outputs, data annotation - get paid $40-$85/hour

Example: Mercor pays 30,000+ contractors $1.5 million/day total to train AI models. Rates: $40-$50/hour for generalist tasks, $85+/hour for specialized work (CPAs, doctors, lawyers). Fully remote. Weekly pay. No commute.

Why it works:

• Mercor alone pays $1.5M/day to 30,000+ contractors (real, verifiable platform)

• Pays $40-$50/hour for entry-level work, $85+ for specialized expertise

• AI labs (OpenAI, Anthropic, etc.) desperately need human feedback to train models

• Work is fully remote and asynchronous (work whenever you want)

• Weekly pay via Stripe or Wise

• No experience required for basic tasks (rating AI responses, data annotation)

• Specialists earn more ($300/hour for expert CPAs training accounting models)

• Platform evaluates 468,000+ applicants, matches skills to projects

• Work trial process - get paid to prove you can do the work

• India and US are biggest talent sources (global opportunities)

Time investment: Flexible - work 5-40 hours/week depending on project availability

Potential income: $800-$3,400/month working 20 hours/week at $30-$85/hour

Difficulty: Beginner-Intermediate (need to pass AI interview, follow rubrics carefully)

Startup cost: $0 (just need computer + internet)

Where I found it: Mercor TechCrunch coverage showing $1.5M/day paid to 30,000 contractors, platform reviews showing $40-$85/hour rates, $75M ARR company working with top 5 AI labs including OpenAI

Tools you'd need:

• Computer (any laptop will do)

• Reliable internet connection

• Payment account (Stripe or Wise for receiving payments)

• Resume (for profile creation)

• Professional skills documentation (if applying for specialized roles)

The catch:

• Work is inconsistent ("feast or famine" - projects pause without warning)

• Many applicants report being ghosted after AI interview

• No feedback if rejected (AI interview is a black box)

• Contract-based (no benefits, no job security)

• Some platforms require invasive tracking software (screenshots, camera access)

• Privacy concerns (you're training AI that might replace your job someday)

• High competition (468,000 applicants evaluated, not all get work)

• Payment depends on approved hours (client can reject your work)

• Specialized roles require verification (degrees, certifications)

My take:

This is the most straightforward "get paid for your brain" opportunity I've seen.

The model is simple: AI labs need humans to train AI. They hire platforms like Mercor to find experts. You sign up, pass an AI interview, complete tasks, get paid weekly.

What kind of work?

• Rating AI responses on a rubric (Which answer is better? Why?)

• Grading model outputs (Is this factually correct? Well-written?)

• Data annotation (Label this image, categorize this text)

• Subject matter review (Does this medical advice make sense? Is this legal explanation accurate?)

For generalists: $30-$50/hour doing basic evaluation work

For specialists: $85-$300/hour training models in your expertise (accounting, medicine, law, finance)

The controversy: You're literally training AI to do your job. A CPA earning $300/hour training an accounting model is teaching AI to replace entry-level accountants. Mercor's CEO calls it the "Reinforcement Learning Economy" - humans get paid to train their replacements.

Is that ethical? Debatable. Is it profitable right now? Absolutely.

Money math:

Conservative (10 hours/week at $40/hour, inconsistent projects):

• Entry-level tasks (data annotation, basic AI grading)

• Projects available 50% of the time

• Income: $800/month (10 hrs/week × 4 weeks × $40 × 50% availability)

• Costs: $0

• Net: $800/month

• Reality: Side hustle income, not full-time

Moderate (20 hours/week at $50/hour, steady projects):

• Generalist AI training work

• Good profile, consistent performance

• Income: $4,000/month (20 hrs/week × 4 weeks × $50)

• Costs: $0

• Net: $4,000/month

• Reality: Livable income in many countries, supplement in high-cost areas

Aggressive (30 hours/week at $85/hour, specialized expert):

• CPA, doctor, lawyer, or PhD-level work

• Training specialized AI models

• Income: $10,200/month (30 hrs/week × 4 weeks × $85)

• Costs: $0

• Net: $10,200/month

• Reality: Competitive with full-time professional work, fully remote

If you want to explore this:

Week 1: Sign Up + Complete Profile

1. Go to work.mercor.com or similar platforms (Scale AI, Appen, Outlier.ai)

2. Sign up with Google or email

3. Complete profile: resume, skills, location, availability

4. Be honest about expertise (specialized roles require verification)

5. Upload professional credentials if you have them (degrees, certifications)

Week 1: AI Interview

1. Platform will invite you to complete 20-minute AI interview

2. Questions are job-specific (coding for engineers, case studies for consultants)

3. You can retake up to 3 times to improve answers

4. Be thorough - this determines your matching

5. No feedback if you fail (frustrating but common)

Week 3-4: Wait for Matching

1. Platform matches your skills to client projects

2. Can take days to weeks to get first project

3. Many people get ghosted here (no response = no match yet)

4. Keep checking email and platform dashboard

5. Consider signing up on multiple platforms (don't put all eggs in one basket)

Month 2+: Work + Get Paid

• Accept projects when offered (they can disappear fast)

• Follow rubrics carefully (quality matters for future matches)

• Track your hours honestly (platforms audit this)

• Submit work on time

• Get paid weekly via Stripe/Wise

• Repeat

Platforms to try:

Mercor (work.mercor.com):

• Best for: Tech professionals, specialists

• Pay: $40-$300/hour depending on expertise

• Focus: Training AI models for top AI labs

Scale AI:

• Best for: Data annotation, labeling

• Pay: $15-$40/hour

• Focus: Computer vision, NLP training data

Outlier.ai:

• Best for: Writers, coders, generalists

• Pay: $20-$50/hour

• Focus: RLHF (Reinforcement Learning from Human Feedback)

Appen:

• Best for: Beginners, global workers

• Pay: $10-$30/hour

• Focus: Data collection, annotation

Common mistakes:

• Rushing the AI interview (take your time, be thoughtful)

• Lying about expertise (platforms verify credentials for specialized roles)

• Poor-quality work on first project (ruins future matching)

• Only signing up on one platform (diversify to avoid dry spells)

• Expecting full-time hours immediately (projects are inconsistent)

• Not reading rubrics carefully (leads to rejected work)

• Giving up after being ghosted (keep checking, projects come in waves)

Red flags this isn't for you:

• You need guaranteed full-time hours

• You hate following detailed instructions/rubrics

• You're uncomfortable with AI interview process

• You need benefits (health insurance, PTO, retirement)

• You're not willing to track hours accurately

• You're paranoid about privacy (some platforms require intrusive monitoring)

• You need immediate income (matching takes time)

Pro tips:

Maximize your matching:

• Fill out profile completely (more details = better matches)

• Upload all relevant credentials upfront

• Be specific about skills (don't just say "writer" - say "medical writer with 5 years pharma experience")

• Update availability honestly (if you can only do 10 hours/week, say that)

Ace the AI interview:

• Read the question carefully

• Structure your answer (intro, body, conclusion for written responses)

• Show your work (explain reasoning, don't just give answer)

• Use specific examples from your experience

• Take all 20 minutes if needed (thoroughness beats speed)

Keep work flowing:

• Sign up on 3-5 platforms (Mercor, Scale, Outlier, Appen, others)

• Check dashboards daily for new projects

• Accept projects quickly (they fill up fast)

• Deliver quality work on time (builds reputation in system)

• Ask for feedback when possible (rare but helpful)

Protect yourself:

• Read privacy policy before signing up

• Use work computer if platform requires monitoring software

• Track your own hours separately (don't rely only on platform)

• Save all work (in case of disputes)

• Know your rights as contractor in your country

The specialist advantage:

If you're a CPA, doctor, lawyer, PhD, or expert in a technical field, you can make serious money.

Example: Mercor reportedly pays expert CPAs $300/hour to train accounting models. Work 20 hours/week = $24,000/month.

Why so high? Because AI labs need experts to train models to expert-level performance. A junior accountant can't train an AI to do complex tax work. You need actual CPAs.

Same for medicine, law, finance, science. The more specialized your expertise, the more you can charge.

The "training your replacement" paradox:

Here's the uncomfortable truth: you're being paid to train AI that will eventually automate jobs like yours.

Mercor's model is explicitly this: hire 100 expert CPAs at $300/hour, train model on accounting tasks until it's as good as an average CPA, now you've automated accounting work.

Is this worth it? Depends on your perspective:

• Short-term: Make $300/hour while the work exists

• Long-term: You're accelerating your own obsolescence

My take: The work exists right now. AI will advance regardless of whether you participate. Might as well get paid while it lasts.

Reality check:

Mercor pays $1.5 million per day to 30,000+ contractors. That's real money going to real people.

But not everyone gets matched. 468,000 applicants evaluated doesn't mean 468,000 people working.

The work is inconsistent. Projects pause without warning. You might work 40 hours one week, 0 hours the next.

This is NOT a stable W-2 job. It's contract work in the gig economy. Treat it accordingly.

But for people who:

• Have expertise AI labs need

• Can handle inconsistency

• Want fully remote, flexible work

• Are comfortable with $40-$85/hour rates

This is one of the cleanest "exchange time for money" plays available right now.

Next issue: A remote cleaning business.

Talk soon, Kris

P.S. - The irony isn't lost on me: AI companies need humans to train AI because AI isn't good enough yet. But the entire point of training is to make AI good enough to not need humans. So you're being paid well today to make yourself unnecessary tomorrow. Take the money, save aggressively, and have a plan for when the work dries up. This won't last forever.

Become An AI Expert In Just 5 Minutes

If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.

This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.

Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.