IBM just developed an AI debater that can reportedly hold its ground while debating with humans.
IBM’s latest program, dubbed Project Debater, made its public debut on Monday against Israeli student debating champions Dan Zafriri and Noa Ovadia. The AI debater, integrated into a black IBM computer, was presented in San Francisco for the first time by the company.
During the Debater’s first face-off with Noa Ovadia, it failed to defend its stand of favoring the government’s plan to subsidize space exploration.
Ovadia was declared the winner in the round by the crowd of journalists present at the event. Even the artificial intelligence system’s attempt at humor didn’t sit well with the audience.
However, the IBM machine learning computer fared better during the second round against champion debater Zafrir. The human and machine argued if telemedicine is indeed worth pursuing.
While Zafrir still won the round, this time it was only by a slight margin. IBM’s Debater was reportedly a winner when it came to knowledge enrichment and was able to persuade at least nine audience members to change their minds and side with it.
Both the AI debater and the human contenders did not have any previous knowledge or experience on the debate topics. Each side was given four minutes to make their opening statement. It was then followed by a four-minute rebuttal and a two-minute closing remark. IBM’s machine learning system went first during the two debates.
According to IBM, the Debater utilizes the IBM Cloud to scan through billions of sentences, allowing it to generate coherent and persuasive positions on different topics. The computer also listens carefully to what its opponent says and formulates a spontaneous and compelling rebuttal, something that has not been done by any machine until now.
“We believe that mastering language is a fundamental frontier that AI has to cross,” Arvind Krishna, IBM’s Research Director, said in a statement.
“There are aspects like speech recognition, speech to text, that AI already does and does quite well. But that is not the same as listening comprehension or constructing a speech that can either be spoken or written or understanding the nuances of claims, meaning what supports a proposition or what may be against a proposition.”
Krishna believes that AI systems that can debate could one day be used by legislators in preparing for debates in critical issues or by lawyers for their briefings.