If you can move cleanly between theory and real-world mess, and you’re the type who can explain an identification strategy without turning it into a TED Talk, this one fits. You’ll challenge AI models on economic reasoning so they stop “sounding smart” and start being right.
About Invisible Agency
Invisible runs large-scale operations and talent programs for fast-growing companies. This AI Trainer project brings in subject-matter experts to evaluate model outputs, document errors, and strengthen accuracy and reasoning for professional-grade use.
Schedule
Remote (United States)
Contract / freelance
Mid-Senior level scope
Hours based on your weekly availability
You supply a secure computer + high-speed internet
No company-sponsored benefits (contractor role)
What You’ll Do
- Pressure-test AI models with practical and theoretical economics scenarios
- Verify factual accuracy and logical consistency across areas like:
- Micro, macro, and behavioral economics
- Econometrics and causal inference
- Public policy, labor, development, and international trade
- Market behavior, game theory, and pricing strategy
- Inflation, unemployment, growth, forecasting, and modeling
- Capture reproducible failure modes (where the model breaks, why it breaks, how to fix it)
- Propose improvements to prompts and evaluation metrics so performance actually improves
What You Need
- Strong economics foundation with the ability to apply it to real decisions and ambiguous data
- Comfort reviewing models, assumptions, and reasoning (not just “the final answer”)
- Clear written communication and a habit of showing your work
- Master’s or PhD in Economics (or closely related field) preferred
- Bonus signals: publications, research experience, hands-on data analysis projects, and applied modeling work
Benefits
- Pay range: $8–$65/hour (rate based on experience, expertise, and location)
- Flexible contractor work
- High-leverage impact: your feedback directly improves future model performance
Real talk: if your econ is mostly vibes and graphs, this will expose you. If you can actually reason, defend assumptions, and spot sloppy logic fast, you’ll be valuable here.
Happy Hunting,
~Two Chicks…