Project Debater: IBM builds robot that argues with humans

Public debate between AI and humans ends in
draw

One-Minute Read

The robot, called Project Debater, was unveiled yesterday at the
tech giant’s San Francisco office, where it took part in couple of
short debates against two human opponents – Noa Ovadia, the
2016 Israeli debate champion, and in a second debate, Dan Zafrir, a
nationally renowned debater in Israel. 

For each debate, the robot and its rival had to prepare a
four-minute opening statement, followed by a four-minute “rebuttal”
and a two-minute closing summary, The Guardian reports.

The topic for the opening debate was the statement that “we
should subsidise space exploration”, followed by “we should
increase the use of telemedicine”.

According to the BBC, the robot analysed “hundreds of millions” of
newspaper and journal articles to come up with its argument to the
topics, which it had not been prepared for beforehand.

Project Debater’s performance is being hailed as a
“groundbreaking” display of AI, as the robot successfully “spoke,
listened and rebutted” human arguments, the broadcaster
adds. 

However, the robot’s public debut wasn’t without a few
hiccups. 

“Some of the points it made were pretty facile” and others
were “clearly cribbed from articles”, says The Verge.

Still, the robot’s ability to engage with complex topics
in near real-time, bar a few minutes to analyse its rivals’
arguments, is “pretty impressive”, the tech news site
admits. 

Some critics are more concerned about how the technology
could be used in the wrong hands. 

MIT Technology Review’s Will Knight argues that
the AI system could be used to power “more pernicious bots on
social media and beyond”.

That claim was disputed by Noam Slonim, who helped develop the
debating bot. Slonim argues that other
technologies pose a far greater threat than the AI system used in
Project Debater.