The debate began between the first AI machine able to talk to man and the champion of Europe in San Francisco

Artificial intelligence (AI) is a hot topic of debate for years, but at the conference IBM Think 2019 in San Francisco (State of California), the technology itself will be the subject of debate. IBM Project Debater, the first artificial intelligence system (AI) capable of debating with humans on complex issues, will face a first-rate opponent until that date Monday, February 11 at 5:00 pm local time (Tuesday at 1:00 pm GMT) at the Yerba Buena Arts Center.

The opponent of Project Debater is the finalist of the World Debate Championship 2016 and the champion of the European debate 2012, Harish Natarajan.. None of you will prepare the discussion topic in advance. Each party has only 15 minutes to prepare their arguments for and against a certain thesis. To defeat Natarajan, the system must gather relevant facts and opinions, structure them with structured arguments, and communicate their ideas clearly and convincingly.

"It's hard to get ready for a debate against an opponent we know little about, but I imagine that Project Debater will pose a challenge," Natarajan said in anticipation of the competition. "From chess to Starcraft, to Jeopardy, the man to machine challenges ended in machine victory."

The debate is divided into an initial statement of four minutes, a rebuttal of equal length and a two-minute summary. John Donvan, Four-time winner of an Emmy Award and host of the series of debates Intelligence Squared U.S., will moderate the session.

Project Debater's knowledge base consists of some 10,000 million sentences from newspapers and academic publications. In development since 2012, Project Debater is IBM's next big step in the field of AI, after previous advances such as Deep Blue (1996/1997) and Watson in Jeopardy! (2011). The goal of the researchers in charge of the project is to build a system that helps people make informed decisions when the answers are not black or white.

Project Debater's first live debate took place last June in front of a small audience against the Israeli champions. Noa Ovadia and Dan Zafrir. Against Ovadia, the machine had to plead in favor of government subsidies for space research.

"Subsidize space exploration, it's like investing in very good tires", Argued the monolith-shaped machine 1.5 meters high, which would enrich the human spirit, inspire the youngest and would constitute a" very solid investment ", he said, making it even more important than public roads, schools or health services. Ovadia was deemed the winner of the debate by the multitude of journalists who attended the event.

The iBM computer performed better in the second debate against Zafrir, arguing this time for the benefits of telemedicine. The machine quoted sources, satisfied the audience and even joked about the process. In general, the crowd felt that the human was more able to communicate the argument, but the computer came out clearly in the category "enrichment of knowledge". At least nine members of the audience have changed their minds on the subject from the point of view of the computer.

"Humans tend to use more expressive language, more original language"They are bringing their own personal experience to illustrate this point," said Darío Gil, director of IBM Research, "the machine does not live in the real world and does not have a life to profit from."

Unlike machines like Watson, IBM says it innovates by creating a system that addresses deeper human practices such as rhetoric and theoretical badysis. "If you think about it, the rules of the debate are much more ambiguous than those of a board game", said Ranit Aharonov, who is leading the project.

"The art of debate is probably more complicated for a machine than for any of these games," said European champion Natarajan in an article published Sunday.

"The success of the debate is essentially based on three elements:. First, a debater must deal with large amounts of information and develop relevant arguments. Secondly, the debate consists of explaining complex arguments in a clear and structured way. Thirdly, it is necessary to make these arguments important for the public. This requires careful use of language, emotions, rhetoric and concrete examples of real life. While a machine must excel in the first, the last two can be a challenge. However, I am sure that there will soon be a point where the AI ​​will be better than the man in all three cases"added Debater's opponent.