PhD Studentship: Verifying Human-GenAI Alignment via Statistical Methods

Generative AI (GenAI) systems, such as ChatGPT, have become increasingly powerful at generating humanlike responses across various applications. However, ensuring that these systems consistently align with human values and preferences remains a critical challenge, particularly in sensitive domains such as healthcare, education, and governance. Current evaluation methods primarily rely on subjective human judgements or assessments by other AI models, offering no formal guarantees of alignment---raising concerns about trust, especially when these technologies are deployed in real-world scenarios.

This project aims to establish provable guarantees for Human-GenAI-Alignment by integrating statistical methods with adversarial methods. For example, by leveraging PAC methods and conformal prediction, we can develop probabilistic guarantees that quantify uncertainty in human preference alignment while ensuring robustness against adversarial conditions. The ability to mathematically verify AI alignment has profound implications. Consider a healthcare assistant system designed to provide medical guidance. If such a system generates misleading or biased advice, the consequences could be severe. Just as safety-critical industries adhere to rigorous reliability standards, the assistant system should be held to quantifiable human-alignment guarantees.

This funded PhD scholarship is suitable for students with a background in Computer Science, Mathematics. Students with interests in machine learning, deep learning, AI, uncertainty quantification, probabilistic methods are encouraged to apply. For eligible students the studentship will cover Home or International tuition fees plus an annual tax-free stipend of at least £19,237 for 4 years full-time, or pro rata for part-time study.

The studentship will be awarded on the basis of merit. Students who pay international tuition fees are eligible to apply, but should note that the award will only provide payment for part of the international tuition fee (~£24k) and no stipend. International applicants should to be aware that they will have to cover the cost of their student visa, healthcare surcharge and other costs of moving to the UK to do a PhD.

The conditions for eligibility of home fees status are complex and you will need to seek advice if you have moved to or from the UK (or Republic of Ireland) within the past 3 years or have applied for settled status under the EU Settlement Scheme.

Advert information