UKRI AI CDT in Decision Making for Complex Systems

The centre works on three themes. Uncertainty in complex systems develops probabilistic machine learning that quantifies uncertainty so that people and machines can decide together and explore new parameter spaces. Decision-making with humans in the loop models machine learning tasks alongside the people who use them, because users often cannot fully state their requirements, and tackles experimental design from scarce data and shifts between domains. Decision-making for machine learning systems looks at automated ways to fit machine learning components safely and accurately into large scientific operations that produce vast amounts of data.

Across all three, the centre puts weight on interpretable and explainable models, for trust and for ethical and legal reasons. A cross-cutting theme adapts the methods for business with industry partners and the Productivity Institute, which is based in Manchester and has a presence in Cambridge.

All students spend the first year at the University of Manchester, in its Department of Computer Science. Four core units of 15 credits each cover reasoning and learning under uncertainty; machine learning and the physical world; reinforcement learning; and an introduction to responsible and reproducible AI, which combines operationalising ethics, team coding hackathons and a challenge to reproduce the results of a published conference paper. A 30-credit research project teaches students to run and report a technical project, and two elective units come from a long list spanning machine learning, robotics, security, health data science, optimisation and finance.

Years two to four are given to the doctoral project, at Manchester or Cambridge, with an annual written report and viva at the end of each year. Students can use the training of Manchester's Science and Engineering Doctoral Academy, sessions in media, policy and public engagement, placements with industry partners, a journal club, one external summer or winter school a year and an annual CDT conference to which partners are invited.

For students with home fee status, the centre pays tuition fees in full and a tax-free stipend of 31,000 pounds in 2026-27 under the TechExpert pilot, pro rata for part-time study and normally rising with inflation, plus funding for conferences and research. International applicants are encouraged but receive the standard UKRI minimum stipend, which the centre gives as 21,000 pounds for 2026-27. EPSRC rules cap international students at 30 percent of a cohort, and because funding for fees is fixed at the UK level, the centre says it takes on finding other sources for the international difference.

The standard entry requirement is an upper second-class honours degree in a subject directly relevant to the PhD, or any 2:1 plus a master's at merit in a relevant subject, or the international equivalent; other combinations of qualifications and experience may be considered. English requirements are IELTS 6.5 overall, with 6.0 in writing and listening and 5.5 in the other parts, or equivalent TOEFL and PTE scores.