Doctoral Student

I'm looking for a PhD student to explore counterfactual reasoning for AI coding agents.Imagine a Tiresias for AI coding agents, the blind prophet from Homers Odyssey, who could foresee what was yet to come. What if AI coding agents had a little of that foresight?Today's coding agents can already write sophisticated software with minimal supervision. But could we make them better at exploring alternative futures for the code they develop?What if AI coding agents could systematically ask: "𝘞𝘩𝘢𝘵 𝘸𝘰𝘶𝘭𝘥 𝘩𝘢𝘱𝘱𝘦𝘯 𝘪𝘧 𝘐 𝘮𝘢𝘥𝘦 𝘵𝘩𝘪𝘴 𝘤𝘩𝘢𝘯𝘨𝘦? 𝘞𝘩𝘢𝘵 𝘤𝘰𝘶𝘭𝘥 𝘨𝘰 𝘸𝘳𝘰𝘯𝘨? 𝘞𝘩𝘢𝘵 𝘪𝘧 𝘐 𝘵𝘳𝘪𝘦𝘥 𝘢 𝘥𝘪𝘧𝘧𝘦𝘳𝘦𝘯𝘵 𝘢𝘱𝘱𝘳𝘰𝘢𝘤𝘩?"This PhD will explore counterfactual reasoning for AI coding agents, bringing together ideas from AI, formal methods, program analysis, and intelligent fuzzing.The ambition is to enable coding agents to anticipate the consequences of hypothetical software changes, identify potential problems, and make better-informed decisions — without necessarily having to implement and test every possible alternative.One exciting possibility is a Foresight Engine: a counterfactual reasoning layer that could be integrated into existing AI coding agents to improve their capabilities, reliability, and computational efficiency.And how would we achieve this? That's where your ideas and intellectual creativity come in.The project is deliberately open-ended.Eligibility: Home students only (under UKRI/EPSRC criteria, including applicable residency requirements).Applications are reviewed on a rolling basis, so early expressions of interest are encouraged.Project details: https://www.qmul.ac.uk/eecs/phd/phd-studentships/giving-ai-coding-agents-foresight---counterfactual-reasoning-for-software-changes/Interested? Email me at v.klimis@qmul.ac.uk for an informal chat.And please feel free to share this with anyone who might be interested.#PhD #ArtificialIntelligence #SoftwareEngineering #FormalMethods #Fuzzing #AIAgents