Harvard Extension School has been the university's continuing education arm since 1910 and its master's degrees have always worked differently from the rest of Harvard. There is no application in the usual sense. You register for the admission courses as a non degree student, pass them to the required standard, then apply during the semester of your second course with the grades already in hand. For the Data Science and Artificial Intelligence ALM those courses are CSCI 101 Foundations of Data Science and Engineering and CSCI 106 Statistical Data Modeling, both online, both four credits, and you need a B or better in each with a cumulative 3.0.
You get two attempts at the B. The programme assumes you can already write Python functions and are comfortable with calculus, and recommends CSCI 7 and MATH 15 first if you cannot. The degree is 12 courses and 48 credits: the two admission courses, four core courses, four electives, an on campus precapstone and an online capstone. Core examples listed this year are Data Mining, Discovery and Exploration, Dynamic Modeling and Forecasting in Big Data, Ethics, Governance and Laws of Data Science, AI and Creative Systems, Introduction to Natural Language Processing and Data Engineering for Analytics. Electives include Deep Learning, Foundations of Large Language Models, Artificial Intelligence, the Internet of Things and Cybersecurity, and Remote Sensing Data and Applications.
The precapstone is taken in person at Harvard in the three week January session or through Harvard Summer School in summer, where you and an industry partner write the research protocol for your capstone, which you then execute with your team the following semester. For international students the summer route comes with an I-20. The rename deserves a paragraph. Until this academic year this was the Data Science ALM, and the old URL still works. For 2026-27 Harvard Extension renamed it, along with its Digital Media Design degree, to put Artificial Intelligence in the title.
I went looking for what changed underneath and found the same two admission courses, the same four plus four plus two structure and the same electives that already existed, several of which were already AI courses. The AI graduate certificate, four courses for $14,320 in foundations of AI, NLP and machine learning, deep learning and computer vision, and AI ethics and governance, can be earned inside the degree, and that is a sensible way to make the AI claim real for yourself. But I would not choose this programme over its previous incarnation on the strength of the new name. Cost is $3,580 per course for 2026-27, up from $3,440 the year before, which puts the twelve course degree at about $43,000 before housing and meals for the summer on campus option. Admitted degree candidates typically receive some Harvard grant aid each term and can use federal loans, which few of the programmes I have reviewed in this series can say.
Still, this is roughly four times Georgia Tech's OMSA, four times the Texas MSDS, and about half of Berkeley's MIDS, for a degree that carries the Extension Studies qualifier. Whether that qualifier matters is the oldest argument about this school. My experience is that it matters to people who went to Harvard College and to nobody else, but you should know it is on the diploma before you spend the money. The rules are strict in a way that suits serious students and punishes drift. Every course after admission needs a B minus or better, the cumulative GPA must stay at 3.0, repeating a course to raise a grade is generally not allowed, you have five years from the term in which you finish any two degree courses, and courses older than five years at the time of admission do not count.
Two withdrawals are free and further ones count as zeros. Read that list twice if you are planning to do this with a demanding job. My own view is that this is the best designed admissions process in higher education, attached to a good but not exceptional curriculum, at a price that only the Harvard name justifies. The MITx MicroMasters pathway is the smart way in: do the four MIT statistics and machine learning courses for $1,350, apply the credential, take one Harvard course, and save yourself a year. If you take the direct route, treat CSCI 101 and 106 as the interview, because that is exactly what they are.