Showing articles associated with Dimitra (Mimie) Liotsiou
Dr Dimitra (Mimie) Liotsiou is a postdoctoral researcher at the Oxford Internet Institute.
Areas of interest
The impact of online information and misinformation on social and political life around the world. Measuring the social influence of online information and interactions using AI methods from the field of causal inference. Analysing the nature, reach and impact of online misinformation and computational propaganda on online discourse and public opinion, using AI and computational social science methods. Fairness, transparency and accountability of computational and AI systems, including investigating the fairness of online recommendation algorithms, particularly using causal inference methods.
What makes Oxford such a good place to work in AI?
Oxford is a great place to work in AI as it has a long tradition of academic excellence and is one of the world leaders in AI research. Also, it is a great place to pursue interdisciplinary work on AI, and is home to a lively community working in this area, offering many opportunities for fruitful collaboration.
What is the biggest opportunity or challenge in AI?
The biggest opportunity right now is getting AI to progress beyond association to causation. Cause and effect relationships are fundamental to human intelligence, and much of human knowledge and reasoning relies on understanding cause and effect relationships (eg that fire causes smoke and not the other way round; that mud does not cause rain). Recent AI research has generally focused on machine learning methods that rely on associational and probabilistic reasoning, rather than causal reasoning. However, the language of association and probability cannot express simple causal statements such as ‘mud does not cause rain’. Hence, there is now a great challenge and an opportunity to achieve this transition to causation, and to bring together causal reasoning in machine learning, as the importance of causal reasoning in AI is now gaining more and more recognition, and there has been extraordinary progress in the research field of causality in the last decades such that there is now a well-developed and mature formal language and calculus of causation. Getting AI systems to progress towards causation will also have big benefits for addressing another crucial problem in AI: ensuring that AI systems are ethical, fair, accountable and transparent.
More about Mimie
