Tech giant IBM has been working on a debating AI since 2012. Unimaginatively dubbed “Project Debater”, the result was a computer containing an AI system whose sole purpose is to “meaningfully debate humans.”
To test Project Debater’s AI-persuading skills, the computer was pitted against 2016 World Debating Championship winner Harish Natarajan. Despite years of intensive research and development from one of the world’s foremost software developers, IBM’s AI still lost to the human debating champion.
Still, its creation alone is a notable achievement in the field of AI, as this is the first one to date that can come up with its own rebuttals, supportive or contradictory viewpoints, then conclude with a closing argument after taking into account the arguments of a human opponent. To formulate its arguments and rebuttals, this debating AI uses newspaper and magazine articles from its own database — it doesn’t rely on a live connection Internet, to simulate the constraints of humans on the debate floor.
IBM’s Project Debater AI is miles ahead of the sort of AI we’re accustomed to, since it held up reasonably well in a field that can offer more than just one solution. Though it looks like Project Debater still has quite a ways to go before it can win an argument with humans, let alone your mother-in-law.
You can read the full story here: https://edition.cnn.com/2019/02/11/tech/ai-versus-human-ibm-debate/index.html
Photo source: https://www.flickr.com/photos/ibm_research_zurich/with/45955713654/