PhD Position in AI Safety, Robust Inference, and Counterfactual Causal Reasoning at King's College London
Nicola Paoletti, Senior Lecturer at King’s College London, is looking for prospective PhD students to join the Department of Informatics. The planned start is October 2027.
The opening is centered on two broad research directions:
Reliable and Robust Inference for AI Safety and Security
, and
Counterfactual Causal Reasoning for Sequential Decision-Making and AI Systems
. The first project focuses on AI monitoring and decision-making with rigorous probabilistic guarantees under distribution shift, adversarial attacks, limited feedback, and evolving environments. Topics include conformal prediction, risk control, sequential and anytime-valid inference, latent monitoring of language models, adversarial robustness, and applications to safety monitoring, attack detection, routing, diagnosis, control, and verification.
The second project explores how counterfactual and causal reasoning can explain decisions and failures in sequential AI systems such as MDPs, POMDPs, and language-model agents. Possible directions include causal attribution and repair for LLM agents, symbolic abstractions of agent behaviour, latent-level causal explanations, counterfactual analysis of learned world models, and formal verification of counterfactual properties.
Funding is available through competitive faculty PhD scholarships and the King’s-China Scholarship Council (K-CSC) programme for eligible Chinese applicants. The main application deadline is December 2026, while K-CSC applicants may follow a different timeline.
Applicants with interests in artificial intelligence, machine learning, AI safety, conformal prediction, causal inference, counterfactuals, formal methods, reinforcement learning, and LLM agents are encouraged to get in touch early to discuss topic fit. The projects are broad and can be shaped around a candidate’s interests and background.
