CISA Seminars - Chris Reed (University of Dundee)

Title

Managing The Worst Sort of Conversation

Abstract

Though often misattributed to Swift, it was his close friend and fellow Scriblerian satirist, Arbuthnot, who wrote that, “Argument, when usually managed, is the worst sort of conversation.” Things have improved little in the intervening centuries. Despite argument being as vital now as it was then for geopolitics, religion and science, and its role today increasing in importance in everything from corporate decision-making to intelligence analysis, our processes of argument, debate and dialectical decision-making remain largely intuitive, subjective and opaque.

A new area in AI has started to yield intriguing results that offer unique potential for improving matters -- whilst also framing profound philosophical challenges. Argument technology aims to support, enhance, augment and track the arguments that humans employ and express, and draws upon a broad range of AI’s interdisciplinary breadth. First, are the new models of argumentation developed in philosophy and linguistics which in turn enable software for discourse analysis, corpus management, language modelling, and nonmonotonic reasoning. Inference Anchoring Theory, for example, underpins tools employed by over 100,000 users, and has enabled the construction of the first million-word-plus corpora of analysed argumentation resources. Second, are the techniques for automatically recognising and extracting argument structure from naturally occurring discourse. Intriguingly, statistical techniques that have been so successful in other areas of language processing are proving insufficient in the face of the communicative sophistication of argumentation. This new task of argument mining is becoming an exciting proving ground for approaches that involve a hybrid of rule-based and statistical modelling. Current work in collaboration with IBM is demonstrating that rules from classical rhetoric, for example, can act as useful priors for machine learning. Third, are the domains of application, with argument technology underpinning debate analytics to enhance participatory deliberative democracy; pointing out reasoning bias in teams of intelligence analysts; training mediators in intervention strategies; supporting dialogue between mathematicians; and most recently, inculcating the skills required for young people to recognisefake news in a BBC-partnered deployment into every secondary school in the UK. The success of such applications attests to the value that argument technology has in contributing to *un*usual managementof conversation.

Biography

Chris Reed is Professor of Computer Science and Philosophy at the University of Dundee in Scotland, where he heads the Centre for Argument Technology. Chris has been working at the overlap between argumentation theory and artificial intelligence for over twenty years, has won £6m in funding and has over 200 publications in the area. He collaborates with a wide range of partners such as IBM and the BBC, and is also active in public engagement and commercialisation of research, having served as executive director (CTO, CSO and CEO) of three start-up companies, and appearing in TV, radio and print media with a combined audience in excess of 29 million people.