2023 was undoubtedly the year of ‘AI’ – when we were all amazed by the capabilities of ChatGTP and image generators, and stories of existential risks filled the media. This narrative seemed to be pushed by the very people who had invested most in developing Large Language Models, and were interlinked with the ideology of Effective Altruism. The concept of AI Safety rose to the top of the political agenda. However, concern about, and research on how AI could be developed and used responsibly has a long history. In 2023 the EU passed the AI Act, which was conceived well before current Generative AI as a framework for governance of risks and harms, while promoting innovation. Edinburgh University has become a leading centre for research on AI Ethics – a paradigm lead by philosophers, rather than one lead by computer scientists and engineers, and emphasises social risks and harms – rather than civilisation destroying harms. However by the end of 2023, the real life requirements of business for robust systems, oversight by government concerned about weapons, jobs, fundamental rights and international power struggles have moved the focus of AI Safety towards practical concerns. In these talks we will hear about how and why existential risk became such a headline grabbing topic, and understand how some ways we can start the hard work of bridging the responsible AI, AI Ethics and AI safety paradigms.

Speaker Biographies

Dr Atoosa Kasirzadeh

Dr Atoosa Kasirzadeh is a philosopher, mathematician, and systems engineer at the University of Edinburgh. She is an Assistant Professor (Chancellor’s Fellow) in the Philosophy Department, Director of Research at the Centre for Technomoral Futures, and a Research Lead at the Alan Turing Institute. Prior to this, she held research positions at DeepMind and Australian National University. 

She has a PhD in Philosophy of Science and Technology (2021) from the University of Toronto and a PhD in Mathematics (2015) from the Ecole Polytechnique of Montreal. Her current research is focused on ethics, safety, and philosophy of AI (value alignment, interpretability, generative models, recommender systems) and philosophy of science (explanation, prediction, complex systems, automating science). 

Website https://kasirzadeh.org/ 

Dr Vassilis Galanos

Dr Vassilis Galanos is a Research Associate and Teaching Fellow at the Edinburgh College of Art, conducting a risk-based assessment of Generative AI in journalism as part of the BRAID UK initiative and teaching about Technology Futures. Vassilis investigates the historical and sociological underpinnings of interwoven AI and internet technologies, and how expertise and expectations are negotiated in these domains. Vassilis is Associate Editor of Technology Analysis and Strategic Management and prospective Lecturer in Digital Work at the University of Stirling’s School of Management.

Website: https://www.sps.ed.ac.uk/staff/vassilis-galanos