We’ve come a long way since the ancient Greeks gathered to debate fundamental human questions. Today, engineering of argument technology is an emerging discipline in AI research. Scientists at IBM developed an AI-powered, fully autonomous debating system called Project Debater. It’s described as “the first AI system that can debate humans on complex topics. The goal is to help people build persuasive arguments and make well-informed decisions.”
Debating with – not about – AI
A paper published in the journal ‘Nature’ fully describes Project Debater’s architecture and thoroughly evaluates how it operates across a broad variety of debate topics. In addition, speeches by both machine and 3 expert human debaters were assessed on almost 80 different topics by 15 members of a virtual audience. To have a meaningful debate, Project Debater relies on groundbreaking AI technology: data-driven speech writing and delivery, listening comprehension and the modelling of human dilemmas. Amongst its core capabilities, Project Debater is able to automatically generate a whole speech and persuasively deliver it, and to understand a long, spontaneous speech by a human in order to compose a rebuttal. Specifically, Project Debater begins by searching for short pieces of text, drawing from about 10 billion sentences sourced from newspapers and journals. It removes redundant text, chooses the best remaining claims and evidence, and organises these by themes to produce a narrative. It then pieces all the selected arguments together to create and deliver a persuasive speech. Project Debater prepares a rebuttal by listening to and processing an opponent’s response.
No debate – we’ve still got the gift of the gab
Project Debater’s abilities were first showcased to the world in 2019 during a live debate. It went head-to-head with universally recognised debate champion Harish Natarajan on whether preschool should be subsidised. With only 15 minutes to research the topic and prepare for the showdown, the AI mined millions of data sets at lightning-quick speeds. The opening statement and counterarguments were delivered by a robotic female voice. The audience declared the Cambridge and Oxford graduate winner of the debate. Project Debater did manage to hold its own. “However, it is still somewhat inferior on average to the results obtained by expert human debaters,” lead author and IBM researcher Noam Slonim told ‘Scientific American’. “Project Debater is, first and foremost, a tremendous engineering feat,” AI expert Prof. Chris Reed at the University of Dundee’s Centre for Argument Technology in the United Kingdom commented in his critique of the ‘Nature’ paper. “Project Debater is a crucial step in the development of argument technology and in working with arguments as local phenomena. Its successes offer a tantalizing glimpse of how an AI system could work with the web of arguments that humans interpret with such apparent ease. Given the wildfires of fake news, the polarization of public opinion and the ubiquity of lazy reasoning, that ease belies an urgent need for humans to be supported in creating, processing, navigating and sharing complex arguments — support that AI might be able to supply.” Project Debater isn’t able to emulate human reasoning or defeat our complex debating skills. We remain the great masters of oratory – at least for now.
AI, artificial intelligence, debate, debating, Project Debater, speech