With the boom of artificial intelligence, numerous arguments and theories about how machines will come to replace humans have arisen. However, several companies developing artificial intelligence projects have insisted on putting their machines to compete against humans. Although they have sometimes won the game, there have also been cases in which they have been defeated by humans.
The IBM AI Debater, Miss Debater, lost a competition of debates against the champion Harish Natarajan within the framework of the IBM Think 2019 convention. The debate revolved around the subsidies to the preschools, and to each participant was given 15 minutes of preparation for a start of four minutes, another four minutes to refute the opponent’s argument, and a two-minute summary.
In front of an audience of 700 people – among those who stood out mainly among the best students of debate in schools in the Bay Area of San Francisco and by more than a hundred journalists – Miss Debater and Natarajan expressed their arguments about the subject in question. According to the official IBM blog, Miss Debater has a knowledge base of 10 billion prayers taken from newspapers and newspapers.
An agile process, but that still can not compete with humans
Even though Miss Debater – formerly known as Project Debater – acts nimbly in the first stage of the debate, some factors still need to be improved to defeat a human being. IBM explains that, in a debate, the machine has to argue about a subject on which it has never been trained.
Once a brief sentence describing the motion is given, Miss Project searches its database of 10 million sentences to make an argument. Although it can be quick looking for information, selecting the message is a much more complex task.
IBM and Natarajan agree that debates can be more complicated for artificial intelligence since a large emotional component and understanding of human language are at stake. In a publication on his LinkedIn, Natarajan said that a successful debate is composed of three elements.
The first has to do with the search for information for the elaboration of an argument; second, there must be a clear and structured explanation, and, finally, that message must be voiced in a way that clicks with the audience.
This requires the careful use of language, emotions, rhetoric, and examples. While one machine can excel in the first, the last two could be challenging. – Natarajan
For its part, IBM says that the entire process requires a deep understanding of human language, which is not always easy for humans and “certainly quite difficult for computers. “For IBM, conquering the use of human language is one of the most ambitious goals of artificial intelligence,” and there is still a way to go.