How we rank.

Our rankings aren't a consumer survey and they aren't a reputation poll. They're a weighted composite of outcome data, research impact, selectivity, and cost-to-earnings ratio. Full formula below.

The four inputs

Every program receives a score from 0–100 on four independently-measured dimensions. We weight them 35 / 25 / 20 / 20 respectively.

1. Graduate outcomes — 35%

Median first-year salary for graduates of the AI track, weighted by placement rate (percentage of cohort employed in AI or ML roles within six months). Source: the program's published career report or, when unavailable, the most recent IPEDS outcome filing.

2. Faculty research impact — 25%

Publication count from core AI track faculty at top-tier venues in the last five years (NeurIPS, ICML, ICLR, CVPR, ACL, EMNLP). Normalized by faculty headcount so small programs aren't penalized.

3. Admit selectivity — 20%

Acceptance rate, with undergraduate GPA and standardized test score context. We intentionally cap this dimension's contribution so a program can't rank well on selectivity alone.

4. Cost-to-earnings ratio — 20%

Total program cost (tuition plus mandatory fees, adjusted for average financial aid) divided by median first-year salary. Rewards programs that deliver good outcomes without extreme cost.

Eligibility

Programs must be regionally accredited in the United States, have graduated at least two cohorts from their AI track, and publish enough outcome data that we can score them without fabrication. ~40 programs currently meet the bar.

What we don't use

Peer reputation surveys (subjective), alumni donations (irrelevant to students), US News rank (circular), or any data supplied directly by admissions offices that isn't independently verifiable.

When the ranking updates

Annually, every March, after employment outcome data from the prior graduating cohort is released. Mid-year we correct individual program data but don't re-rank.