Research Assistant in Machine Learning - AI Alignment and Reward Hacking (Fixed Term)

Department of Engineering

We are seeking a Postdoctoral Research Assistant to join the Machine Learning Group (http://mlg.eng.cam.ac.uk) in the Department of Engineering, University of Cambridge, UK. This position will contribute to the research programme "AI Safety".

The programme's goal is to investigate the various aspects of AI safety, specifically, reward hacking. Candidate will have an opportunity to investigate the phenomenon of reward hacking in great depth and propose methods to prevent reward hacking from happening. Considering the rapid proliferation of deep learning and reinforcement learning agents, the research is expected to impact multiple areas including but not limited to computer vision, natural language processing, natural language generation, robotics, healthcare and autonomous vehicles.

The Research Assistant will work with Dr. David Kruger and other members of the Cambridge Machine Learning Group (http://mlg.eng.cam.ac.uk/). Key responsibilities include developing code to test out different hypothesis, generate original research ideas, running large scale experiments and writing technical reports/papers to report any or all the findings from the study conducted.

https://www.jobs.cam.ac.uk/job/42032/