UKRI AI CDT in Designing Responsible NLP

The programme is organised around five skills domains that the founders drew up with partners: responsible NLP data and models (inclusive datasets and fairer, less biased models); explainable NLP for users in the world (visualisation and human-readable explanations of data and models); designing for human-NLP partnership (human-in-the-loop and human-AI teams); NLP governance and accountability (copyright, data protection, equalities and emerging AI law); and co-creation and NLP futures (design methods that let diverse members of society shape how NLP is used).

Students choose one domain as their main focus, with a principal supervisor who is expert in it, and widen into one or two others. Every student has at least two supervisors from different disciplines or schools, drawn from informatics, design, linguistics, speech science, psychology, law, philosophy, information science and digital humanities.

All students complete 180 taught credits in the first two years alongside the equivalent of three years of research spread over four. In year one the whole cohort takes Researching Responsible and Trustworthy NLP (20 credits, year-long), an Applied Interdisciplinary Project in which teams answer partner briefs in a four-week block (20 credits), an Individual Exploratory Project with two supervisors that shapes the PhD proposal (60 credits), and at least 10 credits of ethics, data and AI courses. Year two adds Case Studies in Responsible NLP, and a training needs analysis steers each student's remaining foundational and specialist courses.

Years three and four are for research and the thesis. The centre also runs a yearly festival linked to the Edinburgh Science Festival, a summer school, masterclasses and industry days, and offers placements, paid internships and training in start-ups, social enterprise and policy. Programme delivery is in person, including for part-time students, who must live within reasonable reach of Edinburgh.

Every studentship waives tuition fees and pays a stipend, with about 17,000 pounds per student for travel, equipment and materials, 8,000 pounds for placements and extra money for computing. For students starting in September 2026 with home fee status, the stipend is the UKRI rate plus 10,000 pounds a year from TechExpert for all four years; overseas students receive the UKRI rate. The University of Edinburgh covers the difference between home and international fees, but visa, health surcharge and relocation costs are the student's own.

UKRI rules let the centre spend no more than 30 percent of its UKRI funds on international students, though university and partner funding sometimes lets it fund more. Successful applicants usually hold at least a 2:1 or a master's in a relevant subject, but applicants with extensive professional or research experience are welcome. Programming skills and an NLP background are not required, though an application must show understanding of language technologies. There is no age limit.