Research methodology

Workflow

1. Discover a provider. Directories, national registers, university lists and search results are used to find leads only; they never verify a fact.
2. Verify identity and operation on the provider’s own website, including that it publishes current information. Closed providers and discontinued programs are excluded.
3. Confirm the teaching location: the campus or studio where teaching takes place, not an office. Online-only providers have no location and are never counted as present in a state or city.
4. Confirm the program on its own official page.
5. Classify the qualification: the exact award and who awards it, whether the credential is a degree, an academic award, a professional credential or a completion certificate, whether it is credit-bearing, and whether the program is a dedicated AI program, a formally named AI pathway within a broader qualification, or a research degree with an AI pathway.
6. Record the subject only where the official program page establishes a specialization, never for individual modules.
7. Extract delivery, duration and schedule: on campus, hybrid, low residency or fully online, required campus time, full-time or part-time study, and live or recorded teaching.
8. Extract prerequisites: programming, mathematics, prior degree and work experience, each as required, recommended, not required or not stated. A missing statement is recorded as not stated, never read as “beginner friendly”.
9. Extract fees in the original currency with their basis (per term, per year, per credit, whole program), period and fee category. Equipment and materials costs are separate lines where published.
10. Save sources: the page address and a short verbatim quote for each fact, with the date it was checked. A script re-fetches the pages and checks that each quote is still present; facts whose quote cannot be found are withheld or downgraded.
11. Withhold any fact that cannot be confirmed. A withheld fact is shown as not published, not filled in from memory or from a third-party site.
12. Deduplicate and publish after review. Every research step is written to a research log.

Who is included

Universities, colleges and community colleges, technical institutions, specialist AI schools, professional training and bootcamp providers, business schools, established technology companies offering structured education, and recognized online platforms (with the actual teaching provider identified). Eligible programs are standalone AI qualifications, broader qualifications with a formally named AI major or pathway, machine learning programs with substantial documented AI content, research degrees with an official AI pathway, and structured AI courses. Teaching language, ownership and international admissions do not affect eligibility.

Excluded: general computer science degrees with only an optional AI elective, generic data science programs without substantial documented AI or machine learning content, promotional webinars, newsletters and content libraries presented as qualifications, software product demonstrations, providers with no verifiable official program information, and closed programs.

Qualifications

We store the exact official award, the normalized qualification category, the awarding institution where a source names it, whether the program is credit-bearing, and the credential type. A completion certificate is never relabelled as a degree, a two-week course is not a diploma because it issues a certificate, and awards from different education systems are not treated as equivalent. National framework levels are shown only with a source and never rename an award.

Geography

The geographic scope is every country and territory in the ISO 3166 dataset plus the 50 US states and Washington, DC as a separate detailed layer. A persistent research matrix and coverage ledger record every geography, every qualification category and every subject group, with a status for each. A search that found nothing is recorded as an outcome of that search, not as proof that no program exists. We do not claim exhaustive worldwide coverage: the directory lists the providers we have researched and verified so far, and the countries page shows the research status of every country and territory. Online programs are located by where the provider is based; eligibility to enrol depends on the student’s location and is shown where the provider states it.

Costs

The original currency is authoritative. We do not convert currencies, rank by price, or compare a per-credit fee with a whole-program fee. Total program price, annual tuition, per-credit tuition, monthly subscriptions, certificate fees and audit access are different bases and are never merged: a monthly subscription is not the cost of completing a program. Free audit, free access, free learning with a paid certificate, an entirely free credential, financial aid and temporary promotions are labelled separately, and a free trial is not a free program. Where a provider does not publish a price we say “not published”.

Editorial selections: how “best” pages are made

The Best AI Schools pages are unranked editorial selections, not league tables, and each page covers one kind of program: a two-hour beginner course and a two-year master’s degree are never placed in one list. There are no ranking numbers and no invented scores, employment rates, salaries or reviews. Each page states what was considered, the review date and the limitations. Degree selections use these published criteria:

• institutional recognition
• AI curriculum depth
• faculty and research
• projects and capstones
• published industry opportunities
• information transparency

Short-course, bootcamp and executive-education selections use: curriculum detail, prerequisite clarity, instructor information, practical projects, assessment, learning support, transparent pricing and the credential issuer.

A program can only be assessed on evidence that is published on official pages, so a good school that publishes little can be missed. Selections are a starting point for your own research, not a verdict.

Sponsorship is separate

Sponsorship has no influence on inclusion, verification, recognition status, results order or editorial selections. Paid placements are labelled and sit in their own slots; see the sponsorship disclosure.

What we do not do

• We do not invent providers, qualifications, recognition, faculty, tuition, scholarships, testimonials, partnerships, reviews, rankings or research coverage.
• We do not imply that any program guarantees admission, employment or professional certification. Bootcamp and course career claims are the provider’s own; we publish outcome statistics only when the provider documents the metric, period, cohort and definition.
• We do not bypass bot protection or access restrictions. Where an official register or page could not be reached, the record says so.

Limits

Coverage is incomplete and uneven by country. Facts are checked at a point in time and can change afterwards. Quotes are re-checked by script, but a script cannot judge whether a page has been reworded in a way that changes its meaning; editors review failures. If you find an error, see corrections.

Freshness

Fees, intakes and programs change. Each record shows when it was last checked; past deadlines are shown as past and are never moved forward.