PhD Studentship Opportunity 

Trustworthy Human-AI Collaboration based on Reinforcement Learning and Foundation Models 

Supervisory Team: 

Dr. Vahid Yazdanpanah (University of Southampton) 

Dr. Jennifer Williams (University of Southampton) 

Prof. Sebastian Stein (University of Southampton) 

Prof. Yulan He (King’s College London) 

Qualification Type: PhD 

Location: University of Southampton, Highfield Campus 

Funding for: UK and EU Students 

Funding amount: Tuition Fees and a stipend of £18,622 p.a, for up to 3.5 years. 

Hours: Full Time 

Closes: No specific closing time (early applications strongly encouraged) 

Interviews: Ongoing as applications are received 

Starting date: January 2024 


Project Description: In this project, you will explore the development of intelligent agents capable of trustworthy collaboration with human end users. The focus extends beyond technical reliability to include the agents' ability to provide assurances, explanations, and address user feedback and preferences. Leveraging reinforcement learning techniques, you will investigate the potential of foundation (language) models for capturing individual-level preferences as well as collective-level norms and conventions. You will also develop proof-of-concepts in collaborative games such as Hanabi or Overcooked to establish the foundation for real-life applications in areas such as smart mobility, disaster response settings, and the allocation of energy resources (e.g., EV charging stations) to end users in connected vehicular systems. The project aligns closely with the Turing AI Fellowship on Citizen-Centric AI Systems (CCAIS) and builds upon the research directions outlined in the position paper titled "Citizen Centric Multiagent Systems".  

CCAIS Project: The CCAIS project aims to develop AI systems with people at their heart. These citizen-centric AI systems learn the preferences of individual users to provide personalised services and advice in important application areas such as smart transportation, energy and disaster response. To ensure trustworthiness, these systems safeguard privacy by learning and making decisions locally, for example on a user's smart device. They also offer clear explanations for their decisions and involve stakeholders through a continuous feedback loop. You will be part of a supportive and diverse research team. As an example of ongoing impactful work in the project, CCAIS postdoc Dr Elnaz Shafipour has recently been featured in a BCS video series on Net Zero. The project has the strong support of a consortium of industrial stakeholders, including IBM Research, Jaguar Land Rover and Yunex Traffic. These will interact with you regularly to ensure your work addresses real societal challenges. 

Research Group: You will join the Agents, Interaction and Complexity group within the School of Electronics and Computer Science (ECS). We are one of the world's leading groups in multi-agent systems, and you will benefit from our significant networks. For example, we lead the UKRI TAS Hub and the UK Hub on Responsible AI, we are part of the Alan Turing Institute, and we run the MINDS CDT. These centres and networks will provide you with a valuable opportunity to work with researchers and other stakeholders from a wide range of backgrounds.  

Informal Enquiries:  Interested candidate are encouraged to reach out to Vahid Yazdanpanah (, Jennifer Williams (, or Sebastian Stein ( 

Candidate profile: Applicants will normally be expected to have a Distinction in an MSc in Computer Science or a related discipline, or an outstanding First-Class BSc or MEng qualification, but all applications will be considered on merit as appropriate to the individual case. We highly value diversity in our teams, and we particularly encourage women, Black, Asian and minority ethnic (BAME), LGBT+ and disabled applicants to apply for this studentship. ECS is committed to Equality, Diversity and Inclusion, and we hold an Athena SWAN Bronze Award. 

Funding: Funding is available for 3.5 years and includes a tax-free stipend at the standard UKRI rate, full time (UK) PhD tuition fees, and an allowance for research consumables, additional training, conference attendance, etc. Eligibility is based on residency rather than nationality. We are pleased to announce that students from EU countries and other Horizon-associated countries are now also eligible for a full studentship (covering tuition fees and living expenses stipend).

Closing date: Application will be considered as they are being received and should be submitted at the very latest by 30th December 2023 for standard admissions, but later applications may be considered depending on the funds remaining in place. The start time would normally be in late January, but a later start date may be possible. 

How to apply: Applications should be made online. Select programme type (Research), 2023/24, Faculty of Physical Sciences and Engineering, next page select “Trustworthy Human-AI Collaboration” (full-time). In Section 2 of the application form you should insert the name of the supervisor, Vahid Yazdanpanah.  

Applications should include: Curriculum Vitae, two reference letters, degree transcripts to date, and a short research proposal that outlines the research opportunities you plan to focus on (1 side of A4 detailing possible scope, aims, methodology and possible outputs). See the following document for guidelines for preparing a Research Proposal:  

Apply online:  

For further information please contact: