Generalist AI training / RLHF work (DataAnnotation, Outlier, legacy Remotasks)

Rank #169 · Tier C · Priority score 46/100 · Category: Micro-tasks & paid research

Write and rate chatbot responses, rank outputs and fact-check AI answers for model-training contractors. Pay is hourly, starting around $25/hour for general work.

Scores

Metric	Value
Payment probability (6 months)	5/10
Automation after setup	2/10
Setup effort (1 = trivial)	3/10
Ongoing effort (1 = none)	5/10
Income potential	5/10
AI leverage	2/10
Legal in reference countries	100%
Platform pays out in reference countries	55%
Capital required	$0
Time to first dollar	7–60 d
Typical monthly income (low / median / high)	$0 / $400 / $3,000

Realistic expectations

Most people who try this earn around $0–$400 per month in the first year; the high figure of $3,000 is a ceiling for a focused solo operator, not a typical outcome. The chance of receiving at least one real payout within six months is rated 5/10. Expect the first payment after 7–60 d.

How it works

AI labs pay data vendors for human preference data. After you pass a writing and reasoning assessment, projects appear in your dashboard, and you are paid hourly or per task for rating, rewriting or red-teaming LLM output. The work is irregular, and projects can disappear without notice. Vendors usually forbid using AI to produce your answers, so this is human labour. Scale wound Remotasks into Outlier.

Platforms: DataAnnotation, Outlier (Scale AI), Remotasks (legacy)

Resources & business plan

What you need to start

Resource	Amount
Time to launch	3 hours
One-time cash	$0
Recurring fixed cost	$0 / month
Tools	DataAnnotation / Outlier dashboards (free); time-tracking sheet (free)
Skills	strong written English; reasoning and fact-checking; following long guidelines (AI drafting usually forbidden)
Equipment	laptop or desktop; stable internet
Accounts	DataAnnotation account + PayPal; Outlier account + ID and phone from your country; tax form

How the money is made (unit economics)

Metric	Value
Unit	per hour
Typical price per unit (low / median / high)	$15 / $25 / $40
Direct cost per unit	$0
Platform fee	0%
Gross margin (median case)	100%
Human hours per unit after automation	1

DataAnnotation's FAQ lists $25-50/hour for general work; Outlier and non-US rates are often lower, hence the $15 low.

Break-even

Metric	Value
Monthly fixed cost	$0
Units per month to break even	0
Typical months to break even	2
Months to recover startup cash	0

No fixed costs, so every paid unit is profit; break-even equals the time to the first payout. Many applicants are never activated, so break-even for them never arrives.

Market conditions (Nov 2026)

Factor	Assessment
Demand trend	stable
Competition	high
Effect of AI	neutral
Price pressure	stable
Chance of real net profit within 12 months (≥ $100/month where the plan can pay that much)	30%
Typical net profit at month 12 (low / median / high)	$0 / $50 / $2,500

AI labs still buy large volumes of preference data, but spend has shifted from generalists to credentialed experts and synthetic data. After Meta's 2025 investment in Scale AI, several labs pulled work from Scale, causing Outlier project droughts. Winners are fluent writers who pass the assessment and get onto steady projects; the common failure is silent non-activation or a sudden project end. Median month-12 net profit was lowered below $100 so it agrees with a profit likelihood under 50% (most starters never reach $100/month).

Pricing strategy

Rates are set per project; move up by passing specialist qualifications (coding, maths, a second language), which unlock higher-rate projects.

Path to profit

Month 0: apply to DataAnnotation and Outlier, take the assessments carefully (cost $0)
Month 1-2: if activated, $100-500/month at 5-10 hours/week
Month 3-6: take specialist qualifications to reach higher-rate projects: $300-1,000/month
Month 6-12: income depends on project supply; keep two platforms active to ride out droughts

Business-plan sources:

Payout path

Methods: DataAnnotation: PayPal -> bank, Outlier: PayPal, Airtm
Minimum payout: DataAnnotation: no minimum stated; Outlier: not stated
Frequency: DataAnnotation on request; Outlier weekly
Typical delay: 7 days
Notes: DataAnnotation pays via PayPal, delivered within a few days of a request. Outlier states it pays weekly; its PayPal/Airtm methods come from third-party reviews and were not confirmed on an official page.

Full cycle (setup → earn → bank account)

Apply
Pass the assessment
Get onboarded to a project
Do the tasks
Withdraw to PayPal -> bank

What can be automated

Calendar/alert automation for new projects
Earnings logging for tax

Unavoidable manual steps

Pass the assessment
Do the tasks manually
Follow project guidelines
KYC

Legality & compliance

Legal everywhere. These are independent-contractor arrangements, and worker classification is being litigated (reported 2025 class action against DataAnnotation).

Compliance requirements: report income for tax; ID verification; no AI-generated submissions where forbidden
Platform availability: Each project limits which countries it accepts. Outlier cites 50 countries and requires a valid ID and phone from your country of residence. DataAnnotation requires English fluency; its country list is not published (reported as US/UK/CA/AU/NZ/IE). Remotasks cut off several African and South Asian countries in 2024.

Risks

Many applicants are never activated
Sudden project droughts
Accounts deactivated without explanation

Evidence that people get paid

DataAnnotation's official FAQ lists starting rates of $25-50/hour (general) and $40-150+/hour (coding/STEM) paid via PayPal. Outlier's official site says contributors are paid weekly in 50 countries. Activation odds are low, hence paymentProbability 5.

Sources

Tags: ai-training, rlhf, human-only

Generated from data/strategies/micro-tasks.json by npm run build:catalog. Edit the JSON, not this file.