FindAPhD Weekly PhD Newsletter | JOIN NOW FindAPhD Weekly PhD Newsletter | JOIN NOW

Understanding dynamics of AI Safety development through behavioural and network modelling

   Centre for Digital Innovation

This project is no longer listed on and may not be available.

Click here to search for PhD studentship opportunities
  Dr T Han, Dr Alessandro Di Stefano, Prof P Van Schaik  No more applications being accepted  Funded PhD Project (Students Worldwide)

About the Project

The race towards powerful Artificial Intelligence (AI) systems between different development teams (e.g., companies, nations), could lead to serious negative consequences, especially when safety and ethical procedures are overlooked, underestimated, or even ignored. In this context, measures such as incentives can be adopted to ensure that AI development teams comply with a set of mutually agreed standards and principles. The teams may engage in multiple races across different AI domains simultaneously (e.g., self-driving, robotics), where they may have heterogeneous capabilities and distinct strategies, making the design of effective measures extremely challenging. Using methods from game and network theories, this PhD project aims to advance our understanding of the mechanisms and dynamics of multiplex AI development races, unveiling how incentive strategies can be designed to ensure positive outcomes and avoid AI disasters. It will inform market, legislative and regulatory efforts to ensuring safe and ethical AI.


Rapid technological advancements in AI, together with the growing deployment of AI in new application domains such as robotics, face recognition, self-driving cars and genetics, are generating an anxiety which makes companies, nations and regions think they should respond competitively. AI appears to have instigated a race among chip builders, simply because of the requirements and demands it places on the technology. Meanwhile, governments – fearful of missing out – are fuelling investments in AI research and development, resulting in a racing narrative that induces yet further anxiety among stakeholders. Races for supremacy across different AI domains may however have detrimental consequences, since participants in these races may well ignore ethical and safety checks in order to speed up development and market penetration. AI researchers and governance bodies are urging restraint and a more thoughtful consideration of the normative and social impacts of these technological developments, as for example emphasised in the UK National AI strategy. 


To advance our understanding of the dynamics of an AI development race, by addressing the abovementioned challenges and assumptions, and therefore generate new insights for the design of more appropriate and effective governance and regulation policies for AI development. To achieve this goal, we shall combine research on incentives and multiplex network modelling with the dynamical approaches to analyze population-wide dynamics, typically employed in Evolutionary Game Theory. Together, they will enable us to systematically explore how different types of incentives (namely, positive vs. negative, peer vs. institutional, and in combination) might influence safety-compliance behaviours over time, and how those behaviours should be configured to ensure desired global outcomes (i.e., high levels of safety-compliance, possibly including compliance sharing).


This research will be built upon methodologies and outcomes from the proposed Direct of Research (DoS)’s past and ongoing research grants, including the Future of Life Institute AI Safety grant (“Incentives for Safety Agreement Compliance in AI Race”) and the Leverhulme Research Fellowship (“Incentives for Commitment Compliance”), see for example the following key publications: Journal of Artificial Intelligence Research 69 (2020): 881-921, PloS One 16.1 (2021): e0244592. The DoS is also very active in the community of AI Safe research community, being a member of the Future of Life institute community and network of selected mentors for this research area (cf. This will ensure that the outcomes produced from this PhD project will make strong impact on this research community. The co-supervisors, Dr Di Stefano and Prof Van Schaik, have strong track record in the project complementary areas of complex and network sciences,see e.g., PloS one, 10(10), e0140646, and behavioural and applied cognitive psychology, see e.g., Cognition, 212, 104666. 

Contribution to school/university strategy:

This project will contribute to the school growing international reputation in behavioural modelling research of ethical and safety AI, enabling these important research areas to become a Teesside and SCEDT research identity, both nationally and intentionally.

AI technologies can pose significant global risks to our civilization (which can be even existential), if not appropriately developed and regulated. This PhD project will make fundamental contributions that help us better understand the scale and nature of these risks, and how they can be mitigated through suitable governance and regulations. Given the international research community’s extensive engagement with AI ethics and regulation and their emphasis in the UK National AI Strategy, this project is timely, topical and of potentially great practical importance.

Entry requirements:

Applicants should hold or expect to obtain a good honours degree (2:1 or above) in a relevant discipline. A masters level qualification in a relevant discipline is desirable, but not essential, as well as a demonstrable understanding of the research area. Further details of the expected background may appear in the specific project details. International students will be subject to the standard entry criteria relating to English language ability, ATAS clearance and, when relevant, UK visa requirements and procedures.

How to Apply:

Applicants should apply online for this opportunity at:

Please use the Online Application (Funded PHD) application form. When asked to specify funding select “other” and enter ‘RDS’ and the title of the PhD project that you are applying for. You should ensure that you clearly indicate that you are applying for a Funded Studentship and the title of the topic or project on the proposal that you will need to upload when applying. If you would like to apply for more than one project, you will need to complete a further application form and specify the relevant title for each application to a topic or project.

Applications for studentships that do not clearly indicate that the application is for a Funded Studentship and state the title of the project applied for on the proposal may mean that your application may not be considered for the appropriate funding.

For academic enquiries, please contact Professor The Anh Han ([Email Address Removed])  

For administrative enquiries before or when making your application, contact [Email Address Removed].  

Funding Notes

The Fees-Paid PhD studentship will cover all tuition fees for the period of a full-time PhD Registration of up to four years. Successful applicants who are eligible will be able to access the UK Doctoral Loan scheme to support with living costs. The Fully Funded PhD Studentship covers tuition fees for the period of a full-time PhD Registration of up to four years and provide an annual tax-free stipend of £15,000 for three years, subject to satisfactory progress. Applicants who are employed and their employer is interested in funding a PhD, can apply for a Collaborative Studentship.


T. A. Han, L. M. Pereira, F. C. Santos and T. Lenaerts. To Regulate or Not: A Social Dynamics Analysis of an Idealised AI Race. Vol 69, pages 881-921, Journal of Artificial Intelligence Research, 2020.
T. A. Han, L. M. Pereira, T. Lenaerts and F. C. Santos. "Mediating artificial intelligence developments through negative and positive incentives." PloS one 16.1 (2021): e0244592.
T. A. Han, L. M. Pereira, T. Lenaerts. Modelling and Influencing the AI Bidding War: A Research Agenda. AAAI/ACM conference on AI, Ethics and Society, pages 5-11, Honolulu, Hawaii, 2019.
Di Stefano, A., Scatà, M., La Corte, A., Liò, P., Catania, E., Guardo, E., & Pagano, S. (2015). “Quantifying the role of homophily in human cooperation using multiplex evolutionary game theory”. PloS one, 10(10), e0140646.
Di Stefano, A., Scatà, M., Vijayakumar, S., Angione, C., La Corte, A., & Liò, P. (2019). “Social dynamics modeling of chrono-nutrition”. PLoS computational biology, 15(1), e1006714.
Martin, R., Kusev, P., & Van Schaik, P. (2021). Autonomous vehicles: how perspective-taking accessibility alters moral judgments and consumer purchasing behavior. Cognition, 212, 104666.
PhD saved successfully
View saved PhDs