How DegreeOutlook Scores Work

U.S. Department of Education College Scorecard — Graduate earnings (1-year and 5-year post-graduation), median debt at graduation, tuition costs, net price by family income bracket, and institutional characteristics. This is the primary source for school+major-specific financial outcomes.

Bureau of Labor Statistics (BLS) — Occupational Employment and Wage Statistics (OEWS) for salary percentiles (10th through 90th), employment projections (2024–2034) for job growth rates and annual openings. Used for career path analysis and job market sizing.

Federal Reserve Bank of New York — Underemployment rates by major, measuring the share of graduates working in jobs that don't require a college degree. Used in the Expected Value employment model.

OpenAI GPTs-are-GPTs Research — Task-level exposure scores measuring what percentage of each occupation's tasks could be performed or significantly assisted by large language models. This is the primary AI risk input.

Felten et al. AIOE (AI Occupational Exposure) — A complementary academic measure of AI exposure that captures broader automation risk beyond LLMs. Combined with GPT exposure for a composite AI risk score.

The DegreeOutlook Score combines three equally weighted factors:

• AI-Adjusted Earnings — Graduate pay accounting for automation risk. Higher earnings with lower AI exposure score better.
• Job Market Size — Hiring demand and annual openings across career paths mapped to each major. Larger markets with positive growth trends score better.
• Financial Value — A blend of ROI (earnings multiple over tuition) and net financial gain that balances affordable programs and high-earning ones. This prevents the score from exclusively favoring expensive elite programs.

Scores are calculated per school+major combination, so the same major can score differently at different institutions based on their graduates' actual earnings and tuition costs.

Each program page includes three scenarios showing how AI could affect 10-year career outcomes:

• Optimistic — Minimal AI disruption (E = 0.1). Job openings remain near current levels.
• Base Case — Gradual AI adoption (E = 0.3). Moderate reduction in field employment probability.
• Pessimistic — Aggressive AI displacement (E = 0.7). Significant job market contraction in exposed fields.

Our model uses an Expected Value employment framework: AI exposure reduces the probability of landing a job in your target field — it doesn't compress salaries. Nominal wages are sticky; employers cut headcount before cutting pay.

The formula: Expected Earnings = P(field employment) × Target Wage + P(underemployment) × Fallback Wage

Where P(field employment) factors in AI task exposure, BLS projected job growth, and historical underemployment rates for each major.

The "Earnings Multiple" shown on program pages is total 10-year projected earnings divided by total tuition cost. It is not a discounted financial ROI — it's a simple ratio showing how many times over you earn back your education investment.

Published sticker-price tuition is used for cost calculations. Most students pay significantly less after financial aid — see each school's page for net price breakdowns by family income.

College Scorecard earnings data represents median outcomes and may be based on small cohort sizes for some programs. Earnings reflect graduates who filed federal taxes — self-employed individuals and those who left the workforce are underrepresented.

AI exposure scores are based on current LLM capabilities and academic projections. Actual automation timelines are inherently uncertain — our three-scenario model is designed to bracket this uncertainty rather than predict a single outcome.

BLS employment projections are national averages. Regional job markets can differ significantly from national trends.