Hear, Hear: This AI Can Challenge Humans to a Round of Debate

Technology > IT

IBM building in Amsterdam / Photo by Wikimedia Commons


The USA Today reported that IBM showcased an artificial intelligence-enabled computer called Project Debater that can be used in the manly art of debate. To show that such a system is feasible, IBM staged two rounds of debate with humans, with journalists witnessing the verbal skirmish.

The first was with Israeli debating champion Noa Ovadia on the subject of government subsidies for space exploration, with the IBM computer favoring such a policy while Ovadia was against it. While the journalists pronounced Ovadia as the winner of the debate for delivering the more convincing argument, she was outscored by the IBM computer on the subject of knowledge enrichment. 

The second debate was with another Israeli debater, Dan Zafrir, on the subject of telemedicine. While Zafrir was adjudged the winner by journalists by a slim margin based on oratorical delivery, the IBM computer was a big winner again in knowledge enrichment. Furthermore, nine members of the audience changed their opinions on the subject to align with the computer’s viewpoint which was in favor of telemedicine. 

Neither the debaters nor the computer was informed of the topics to be argued in advance. Each side was given four minutes to deliver an opening statement and another four minutes to rebut the other side’s argument, followed by a two-minute closing summary. On both occasions, the computer was given the chance to give its side ahead of the human debaters. 

The event, held in the city of San Francisco, California, was the first time that the public was given an opportunity to witness a debate between a human and a computer. Previously, IBM scientists were debating with the computers on such subjects as the necessity of income taxes, the safety of autonomous cars and if antibiotics should be used in foods.