IBM has developed a robot called Debater, which recently wrestled with a human in a debate over whether the government should subsidize kindergartens. Of course, what you really want to know is whether artificial intelligence (AI) can defeat human beings in the debate.
Compared with IBM's Computer Dark Blue defeating humanity in 1997 and AI's defeating the best chess player in 2017, this human debater's victory may not be so eye-catching. But IBM's robot shows that AI can still be used in ambiguous situations and in debates. It's very different from just scoring to determine who wins and who loses.
Although IBM Debater failed this time, there is no doubt that it is still the winner in some way: you will listen carefully to what it is saying, not just because it is a computer. It divides its arguments into several parts and uses various research data to support them. Although it's not perfect, there's no such thing as a bull's head and a horse's mouth.
Depending on the IBM cloud computing infrastructure, Debater uses a set of powerful machines.
The scoring rule of the game is to see how many people have changed their minds. Before the debate, 79% agreed with the government subsidizing kindergartens, but after watching the debate, the number dropped by 17 percentage points to 62%.
In today's era, Apple's Siri, Amazon's Alexa and Google Assistant all understand our questions well and answer them in voice, so we often forget how amazing it is to be able to talk to a computer. IBM Debater goes a step further and can talk for several minutes at a time.
Should the government subsidize kindergartens?
In this debate, Debater is the right person to support subsidies for kindergartens. Natalayan, the human debater, is the opposite.
The most challenging competition so far
For IBM's debating AI, the game is by far the most difficult.
This is because the technology behind the Debater project focuses on the complexity and nuances of the real world, rather than the kind of non-win or lose in chess games.