University of California, Berkeley

Named groups are the most reliable public signal of where a department's attention actually sits, because they carry funding and faculty rather than only marketing copy. Whether any of them will take you is a separate question, and one worth asking directly before applying.

Recognition

The American system

A US masters in computing normally runs two years, with substantial coursework and room for research assistantships or internships between them. Terminal masters degrees are less central than in the UK: many departments are oriented toward doctoral study, and a masters is often either a stepping stone or a professional qualification rather than the main graduate product. Funding for masters students is scarce, and the cost is high. Application deadlines fall around December for the following September, and most programmes ask for the GRE, references and a statement of purpose.

The programme

EECS graduate programmes, at University of California, Berkeley.

Entry requirements, fees and deadlines are on that page and deliberately not reproduced here. They change every year, cannot be verified at scale, and a figure that is twelve months old is worse than no figure at all.

What to check on the syllabus

No ranking measures teaching quality, so the only way to assess a taught programme is to read what it actually contains. Six things worth looking for, in the published module list rather than the prospectus.

• Statistics and probability as prerequisites, not options. A programme that admits you without them will also teach around them.
• Evaluation as its own subject. Measuring whether a system works is the hardest part of the job and the part most often assumed rather than taught.
• Failure modes and monitoring after deployment. Models degrade and distributions shift, and almost no curriculum covers what happens after a system ships.
• A capstone with real stakeholders, not a benchmark competition. Optimising a leaderboard teaches a skill that does not transfer.
• Some governance, auditing or measurement validity. The regulatory surface around this work is growing faster than the technical one.
• Enough mathematics to read a paper. A programme that teaches frameworks produces someone who can use this year's tools.

Subfields it publishes in

General AI, Machine learning, Computer vision. If you already know which area you want, subfield strength is a better guide than an overall position, since a department can be outstanding in one and ordinary in another. Publication-based rankings can be filtered by subfield; reputation-based ones generally cannot.

Stating the gap is part of the entry. A page that answers every question about an institution is either drawing on sources it has not named or filling in what it does not know.

Before you use any of this

The two serious rankings disagree completely about who leads AI. QS puts MIT first; CSRankings puts Peking University first and finds eighteen of the top twenty in Asia. That is a methodology difference rather than a contradiction, and the explanation, along with the rubric above in full context, is worth reading before choosing on a number.

Sources for this page

Ranking positions from the QS Data Science and Artificial Intelligence table for 2026 and the CSRankings 2025 and 2026 editions. Founding year and institutional history from the reference page below. Programme details from the institution's own admissions pages. Deliberately absent: fees, acceptance rates and entry requirements, which change annually and cannot be verified at scale.

Department and admissionsReference and imagesOnline courses instead

Common questions

Is University of California, Berkeley good for AI?

University of California, Berkeley appears in this list because of its position in at least one of the two serious rankings, or because of a documented departmental strength. It does not hold a verified top-twenty position in either table, and is included for the departmental strengths described above. Neither ranking measures teaching quality, which is what the curriculum rubric is for.

What AI programme does University of California, Berkeley offer?

EECS graduate programmes. The link on this page goes to the institution's own page, which is where entry requirements, fees and application deadlines are current. Those are deliberately not reproduced here, because they change annually and a stale figure is worse than none.

Which areas of AI is University of California, Berkeley strongest in?

Based on the subfields it publishes in: general ai, machine learning, computer vision. Subfield strength matters more than an overall ranking position if you already know what you want to work on, since a department can be outstanding in one area and ordinary in another.

How should I use university rankings when choosing?

Decide which question the ranking answers before you read it. For a PhD, publication output at top venues is the relevant signal and CSRankings measures it directly. For immediate industry hiring, employer reputation matters and QS weights it at 15%. For a taught masters, neither measures curriculum quality, so read the syllabus against the six-point rubric instead.