A match between an Israeli college debate champion and a loquacious IBM computer program demonstrated on Monday shows that computers can hold conversations with humans. Noa Ovadia, a college senior who won an Israeli championship in 2016, squared up with IBM’s program, IBM Debater. She argued against government subsidies for space exploration. The machine argued in favor, delivering three brief speeches. It was as if the program was replying to Ms. Ovadia’s arguments. This artificial intelligence system has been developing for the past six years and is a part of some broader technology build which can interact with people.
Previously companies like Apple and Google have brought to market, gadgets with AI which can do small tasks and answer simple questions for humans. Like last month, Google demonstrated its brand new Google Duplex which can phone a restaurant and make reservations. Projects like these remind us that technology has stretched very far. IBM system’s interactions are tightly constrained: a four-minute opening statement followed by a rebuttal to its opponent’s argument. Noam Slonim helped oversee this project and he estimated that this system could have a meaningful debate on those 100 topics ingrained in its system 40% of the time.
Certainly it has a long way to go to pass the Turing test. Back in 2011, IBM used another system called Watson which beat the leading players at the trivia game show “Jeopardy!” After Watson, this is IBM’s next grand challenge. This system is of course is not build for business, but it reflects the recent acceleration of research related to “natural language understanding”.
As this research progresses, it can provide new ways for computers to digest and process information or even lead to machines that can hold a completely convincing conversation. IBM Debater leans on a wide range of systems and each system will handle different part of the problem of understanding natural language. Each system is designed and built separately before researchers meld them together.
“It is now very obvious this change is happening,” said Jeremy Howard, an independent researcher working in this area. “But these things take time.” We still have a long way to go before the AIs will be able to hold their own grounds in conversations and whatnot.