Nandi Schoots (2019 cohort), C Henrik R Åslund (2019 cohort), Dylan Cope (2019 cohort), Matt MacDermott (2021 cohort) and Francis Rhys Ward (2020 cohort) of the UKRI Centre for Doctoral Training in Safe and Trusted AI, were the organisers of the Safe and Trustworthy AI Workshop 2022, along with Rob Kirk (UCL).
The Safe and Trustworthy AI Workshop took place on 2 November 2022. The aim of the workshop was to make progress on guarantees that AI systems are safe and trustworthy, and brought PhD students from London universities and other early career researchers together for a poster session. In total, there were 39 accepted posters, from researchers across institutions, including the STAI CDT, Oxford, Cambridge, York, Stanford and Google New York.
The workshop was a great success, with over 150 attendees. It also featured two talks from guest speakers, Tom Everitt and Jesse Clifton. Tom Everitt, a Senior Researcher from the DeepMind safety team gave a talk on, ‘Causality in AI risk’. Jesse Clifton, the Lead Researcher at the Centre on Long-term Risk, and a research analyst at the Cooperative AI Foundation, gave a talk on ‘Differential progress in cooperative AI: motivation and measurement’.
Francis Rhys Ward, one of the organisers, said, “On behalf of the organizers I’d like to thank everyone involved in the workshop. We found it to be really enjoyable and hope the attendees did too”.
Prizes were given out for the five best posters. The workshop was supported by the UKRI Centre for Doctoral Training in Safe and Trusted AI; the prize was supported by the Longterm Future Fund.