By: Britt Terrell
IBM Research conducted the first live debate between a human and a machine this week, showing off the company's latest development in artificial intelligence.
Project Debater squared off against two experts in rhetoric Monday night, responding to and delivering its own arguments. The technology, which ingests information from newspaper articles and other sources to put together well-structured speech and deliver a response, was even able to convince some audience members to change their minds.
"This is a great tool for helping educate the public," said Arvind Krishna, SVP & Director of IBM Research. "They could ask it a question, and 'Debater' can then look through and say, 'Look, based on everything I've read, here is an informed opinion on what everybody's thinking and saying.'"
When it comes to using the machine to help humans formulate opinions, though, especially in an age when fake news has run rampant on certain sites, Debater's powers could raises some concerns. But Krishna pointed out that humans are subject to the same pitfalls.
"What the Debater will do is it won't stick only to the fake news, it will give other [opinions], so you really get a complete and comprehensive [response]."
Debater is not the first example of high-functioning A.I. to come out of IBM. The company's Watson technology showed off against Jeopardy! champions in 2011, and its 'Deep Blue' program famously bested chess champion Garry Kasparov back in 1997.
But there is still work to be done on the Debater software. And while Krishna dismissed criticisms that its voice was too robotic ー Google, after all, drew jabs for its all-too-human-sounding Duplex ー he acknowledged some room for improvement.
"How do you throw humor in, when do you throw humor in?" he asked. "Like when it makes a joke like, 'If I had blood, my blood would boil, but I don't really have blood.'"
For the full interview, click here.