IBM was not satisfied with making artificial intelligence (AI) that can defeat world-class chess masters and Go players. An article on Venture Beat reported that the American computer company has just debuted an AI that can win arguments against skilled debaters.
Project Debater is a prototype AI that is capable of two-way conversation. It can crack jokes, is rather rude, and occasionally will leave professional debaters speechless.
Its creator, Noam Slonim of IBM Research, explained that his laboratory completed building the AI in 2012. It has spent all those years training itself on news articles and journals. It was considered ready for real debates by 2016.
Upon encountering an unfamiliar subject, Project Debater browses its database for the right lines. It picks them according to applicability, combativeness, and usefulness in its argument.
It will try to interpret the primary theme in a debate before setting up its argument. (Related: AI robot can draw what you’re thinking by reading your brain impulses.)
The AI recently demonstrated its persuasiveness in an engagement with two professional Israeli debaters. They argued over whether or not space exploration should be subsidized by national governments, as well as if healthcare should give a larger role to telemedicine practices.
Each debate was organized into an opening speech, a rebuttal, and an argument summary. Project Debater spent the time in between each segment to process its opponent’s words. The only scripted parts were the greetings and jokes.
Debate is a much more subtle contest than a game of chess or Go. Still, the IBM artificial intelligence managed to sway many attendees to its side. The results of the debate showed that humans are better at giving speeches while the AI was better at providing information to the audience.
In the telemedicine debate, Project Debater defeated its human opponent by persuading nine out of the 40 audience members. The AI could quote many different sources during both debates.
To avoid misquotes or problems with its text-to-speech translator, it did not directly cite its opponents. It still made some mistakes like claiming that space exploration was more important than roads and randomly measuring words.
Director Chris Reed of the Center for Argument Technology was one of the observers during Project Debater’s public performance. He noted that the AI did not forget its stance, had perfect grammar, and even predicted the arguments of its human opponents. The last ability allowed Debater to counter the humans’ arguments before they aired them out.
“Argument and debate — essentially, it’s the engine that drives the process of science, characterizes what happens in most political forums, and even frames most conceptions of modern religion,” remarked Reed. “Argumentation is one of the defining features of what it is to be human, and if we can convey part of that, then I think that means something very important is starting to change.”
IBM researcher Slonim explained that his team needed to develop many new techniques before they could make Project Debater a reality. Once they completed their work on it, the AI demonstrated several significant advantages over human counterparts.
The biggest buff has to be its huge database of published articles for materials. The AI can easily access this database for material that supports its position and refute arguments, whereas humans have to contend with slower memories and far more restricted knowledge.
Project Debater will undergo further testing tested at an extended debate within the latter half of 2018. Slonim added that IBM would organize a workshop after its next debate so that debate authorities can evaluate his AI’s performance.
Don’t lose track of these frighteningly argumentative AI systems at Robots.news.