In light of recent advances in AI capabilities there has been a great deal of action from governance stakeholders – including multinational institutions, national governments, and the private sector – aiming to address the potential risks of AI, while capitalising on its benefits. However, there remain significant gaps between policy desiderata and the realities of what is technically feasible. Furthermore, decision-makers often have insufficient technical information for assessing the relative efficacy of different policy options.
The emerging field of technical AI governance – that is, technical analysis and tools for supporting the effective governance of AI – aims to address these gaps. This can be through identifying areas where governance intervention is needed, assessing the efficacy of potential governance actions, and enhancing their implementation by providing technical mechanisms for enforcement, incentivisation, or compliance.
The Oxford Martin AI Governance Initiative aims to advance technical AI governance through conducting in-house research, funding DPhil students in the computer science and engineering departments working on technical governance topics, as well as facilitating connections and collaborations between technical and policy experts, both at Oxford and beyond.
