As you know, in a dispute, truth is born. But this happens only if the parties agree with each other's arguments, without turning the dialogue into a farce. And if emotions can prevail in a dispute between people, then emotions are alien to machines. All the more interesting is the fact that in the United States in San Francisco not so long ago the first dispute between man and AI took place. Moreover, the latter, defending his position, relied on facts and presented various arguments.
The new AI has been developed by IBM since 2011 and bears the self-explanatory name Project Debater. Moreover, interestingly, for 5 years Project Debater only discussed with other AIs, and first entered into a skirmish with a human only in 2016. Noam Slonim, one of the project leaders and member of IBM Research, said
"Whenever a topic is raised in a discussion, artificial intelligence scans a huge array of scientific articles and media data for the right assumptions and arguments that will help support its point of view and the discussion itself."
During the debate in San Francisco, Israeli presenter Dan Zafir acted as an opponent of AI. The topic of discussion was the following: "Should the state provide subsidies for space exploration and telemedicine." Project Debater cited a range of sources, from the words of the Sheikh of the UAE (United Arab Emirates) to official statistics from the German Ministry of Economy. AI also raised issues related to the availability of jobs in the areas discussed. The audience observing the process decided that in one of the rounds of the debate, the AI had prevailed over the human. However, the AI for debate is still imperfect. For example, according to eyewitnesses, he sometimes made completely incoherent arguments and never quoted an opponent during an argument, which in real debates often brings a positive result.
Vladimir Kuznetsov
