Virtual Talk with Carina Prunkl
The human mind, while extraordinary, is prone to fluctuations and biases in judgment, influenced by factors such as fatigue, hunger, and inherent predispositions. These inconsistencies, often unwelcome, have been termed "noise" by Kahneman et al. (2021). Such 'noise' or unwanted variability in judgment is commonplace, influencing decisions ranging from everyday choices to courtroom sentences. Algorithms offer a promising counterpoint to these inconsistencies. Unaffected by the fatigue, hunger, or cognitive biases, algorithms are much more consistent in their decision-making. But can these digital tools truly help us overcome noise in human judgments? Kahneman et al. argue that they can, if not eliminate, at least significantly reduce noise. In this talk, I will dissect the notion of 'noise' in detail and scrutinise the claim that algorithms are noise-free.
Dr Carina Prunkl is a Research Fellow at the University of Oxford's Institute for Ethics in AI, a Junior Research Fellow at Jesus College, Oxford. Her current research focuses on automated decision-making, algorithmic fairness, and human autonomy, as well as policy and community governance of AI.
Previously, Dr Prunkl was a Senior Research Fellow at the Future of Humanity Institute and a Visiting Postdoctoral Fellow at Harvard University's Black Hole Initiative. She's also worked as an Ethics consultant for Digital Catapult and the AI Lab at Vrije Universiteit Brussel. Over the years, Dr Prunkl has engaged in a wide range of collaborations, including with OpenAI, the Partnership on AI, and the Centre for the Studies of Existential Risks. Further engagements include the UK government's Centre for Data Ethics and Innovation, the UK Ministry of Defence, the EU Delegation in Russia, the Mexican Senate, and the UK 2070 Commission. Dr Prunkl has designed and delivered multiple courses on the Ethics and Governance of AI for undergraduates and graduates in Philosophy, Engineering, and Computer Science at the University of Oxford.
Please register with email@example.com to receive the Zoom login.