AIDegrees ranks programs using independently collected data. Our rankings emphasize affordability, student outcomes, and the strength of each program’s AI curriculum. We use publicly available federal data and information published by colleges and universities. Factors and weights vary when student priorities materially differ—for example, affordability receives greater weight in online-program rankings, while research strength receives greater weight in doctoral rankings.
Key facts
- Placement is never sold. Schools cannot pay to be included or ranked higher.
- Directories help us discover programs. They do not determine rank order.
- Most Affordable lists are 100% estimated program tuition. They are not a composite score.
How we rank Best Overall lists
On “best” lists, each eligible program receives a 100-point score. Numeric inputs are converted to percentiles within the programs on that page, and within the same degree level so a bachelor’s program is not scored against a master’s on earnings or tuition. Cost and debt use reversed percentiles (lower is better). Extreme values are winsorized at about the 5th and 95th percentiles. Displayed scores round to one decimal place.
The default bachelor’s model is:
| Component | Weight | What it answers |
|---|---|---|
| Affordability | 30% | How financially accessible is this program? |
| Career outcomes | 25% | How do similar-field graduates fare on earnings and debt? |
| AI curriculum | 25% | How substantial and specific is the AI coursework? |
| Experiential learning | 10% | Is there a final project, internship, thesis, or substantial project? |
| Research environment | 10% | How strong is the surrounding research classification? |
Outcomes split 20/30 earnings and 10/30 debt when both exist. If one is missing, the available measure carries the outcomes weight. Missing earnings or debt are never scored as zero. If we cannot estimate tuition, the program receives no affordability points so unpublished prices cannot outrank priced programs.
Weights by ranking type
| Ranking | Affordability | Outcomes | Curriculum | Experiential | Research | NSA CAE |
|---|---|---|---|---|---|---|
| Best AI bachelor's | 30% | 25% | 25% | 10% | 10% | — |
| Best online AI bachelor's | 40% | 30% | 20% | 10% | — | — |
| Best AI master's | 30% | 30% | 20% | 5% | 15% | — |
| Best online AI master's | 40% | 30% | 20% | 10% | — | — |
| Best AI PhD | 10% | 15% | 20% | 10% | 45% | — |
| Best value | 50% | 40% | 8% | 2% | — | — |
| Best cybersecurity (campus) | 35% | 30% | 15% | 5% | 5% | 10% |
| Best online cybersecurity | 40% | 30% | 18% | 7% | — | 5% |
| Best cybersecurity PhD | 10% | 15% | 18% | 7% | 35% | 15% |
| Certificates / short programs | 45% | 25% | 25% | 5% | — | — |
Most Affordable pages do not use this table. They sort solely by estimated program tuition, lowest to highest, and omit programs without a defensible total.
How we estimate tuition
- Prefer a published program total.
- Otherwise multiply required credits by the applicable tuition rate.
- Use the online rate for online programs when it is labeled.
- Keep in-state and out-of-state prices distinct. We never take the lowest unlabeled dollar in a mixed string.
- We do not substitute institutional net price for graduate tuition.
Student outcomes
When a school matches a U.S. Department of Education institution record, we use field-of-study median earnings and cumulative debt for related computing and data fields at the same credential level. We do not use whole-school earnings as a stand-in for an AI program. Privacy-suppressed or missing fields are skipped and remaining outcome weights are rescaled. Each ranking page reports data coverage for the list. Institution and field-of-study files come from the Department’s data release dated June 10, 2026.
AI curriculum rubric (25 points scaled to the page weight)
| Measure | Points |
|---|---|
| AI curriculum breadth (AI, ML, deep learning, NLP, vision, robotics, generative AI, RL, ethics) | 10 |
| Advanced AI depth (survey vs concentration vs dedicated curriculum) | 6 |
| Program specificity (dedicated AI degree vs concentration vs related) | 3 |
| Curriculum transparency (verifiable program page, credits, tuition, format) | 2 |
We score those measures from course titles on the school’s official curriculum or program page when we can extract a course list. Marketing copy and competitor blurbs do not count. A mention of “AI” in a course description is not enough — the course title has to name the topic. If we cannot harvest a course list, the program keeps a modest title-only score so missing course lists are not treated as zero. Experiential learning is scored separately from final project, practicum, thesis, internship/co-op, or a substantial project course.
Research environment
Version 1 uses Carnegie bands from the U.S. Department of Education institution file: R1 = 100, R2 = 75, Research College/University = 50, other = 25. Online “best” lists set research weight to zero so research prestige cannot dominate a working-adult ranking. Doctoral lists raise the research weight. When tuition and field-of-study outcomes are sparse on PhD pages, order largely reflects Carnegie research bands plus curriculum clues—not a full affordability or earnings sort. We do not apply a “world-class quality” discount to high tuition.
Cybersecurity rankings
Cyber lists use the same engine with a smaller curriculum weight and an NSA Center of Academic Excellence factor drawn from the official CAE Community institution map (not directory marketing copy). That is a program-quality marker, not a substitute for curriculum review. Related CIP awards from NCES IPEDS Completions (C2024_A) are attached as membership metadata when a UNITID matches—they do not currently change the ranking method. We also seed program membership from IPEDS CIP awarders missing from competitor directories (especially cybersecurity CIP 11.10 and AI/data CIP 30.70/30.71); those seeds still need school program pages and tuition before they can lead affordability-heavy lists.
Best Overall vs Best by Cost
Best (or Best Overall) pages use our ranking method—a weighted mix of affordability, outcomes, curriculum, and related factors for that ranking type. Most Affordable (Best by Cost) pages ignore the composite score and sort solely by estimated program tuition, lowest to highest. Use Best when you want a balanced shortlist; use Most Affordable when published list price is the main filter—then still check curriculum fit and accreditation.
Eligibility
A keyword in a scraped row is not enough. For an AI ranking we require a named U.S. program, the correct degree level and delivery mode for the page, and material AI/ML (or cybersecurity) representation in the program title or topic. Affordability rankings also require a defensible tuition total.
What we do not score
- Acceptance rate, SAT/ACT, or “admissions accessibility”
- Overall school rankings or generic prestige
- Faculty/student ratio or institutional enrollment
- How many programs appear
Discovery vs ranking
Public program directories, school sites, and federal completions data help us find candidates. Official program pages, published tuition, U.S. Department of Education field-of-study and institution files, and Carnegie classification feed the score. Discovery sources never determine where a program ranks.
Update cadence
We refresh major rankings when program data, tuition harvest, or federal files update. Confirm current tuition and admissions on each school site before you apply.
How rankings are produced: AI + machine, with human oversight
The numbers behind every ranking come from software, not from hand-typed estimates. Python scripts collect program listings and federal data, standardize school names and tuition, compute each score with the weights above, and run automated checks (for example, that Most Affordable lists really run from lowest to highest cost) before pages are rebuilt. AI writing tools help draft and edit the text around those numbers, such as intros and program summaries. People set the methodology and weights, review the output, decide what is published, and correct errors readers report. More on how this site is made.
Editorial independence
We do not sell ranking placement. Affiliate or partner relationships—if any—will be disclosed on the relevant page and never determine rank order. Corrections go through contact. See our editorial standards for how we source and check figures.