I saw the future of artificial intelligence, and it was both thrilling and terrifying.
IBM (IBM) on Monday evening demonstrated Project Debater, the first A.I. system capable of intelligently sparring with a human being through debate. Project Debater, which runs off a group of computers in a remote data center, is capable of pulling information from over 300 million indexed news articles and academic papers, listening to its human opponent and forming a coherent argument, including an opening speech, rebuttal and closing summary — all spoken by an upbeat-sounding woman’s voice.
At its event, IBM demonstrated the Project Debater’s abilities over the course of two sample debates, one on subsidizing space exploration, the other on telemedicine. At times, it proved extremely impressive, such as its opening argument on the benefits of telemedicine — a multi-faceted opener that quoted stats and figures from multiple sources on a topic it had just been provided minutes before.
Project Debater even showed off some tongue-in-cheek humor.
“I can’t say it makes my blood boil, because I have no blood, but it seems some people naturally suspect technology because it’s new,” the Project Debater remarked during its closing argument on telemedicine.
The ushering in of a frightening new era?
Still, there was the occasional rough edge hinting at Project Debater’s limitations in its current form. The research that emerges from space exploration, it argued, far outstripped the costs of government spending on such ventures, even comparing subsidizing space exploration to investing in “really good tires.” It also said such investments were more valuable than good roads, schools or health care — a questionable statement of logic that elicited light chuckles from the audience.
But Project Debater’s abilities were best summed up when it displayed a hint of self-awareness. When debating for space exploration subsidies, it said it supports technology because “I am a prime example of its power.”
In a room full of forward-thinking Silicon Valley journalists, the statement was harmless enough. But the remark might not have gone over as well in areas and industries feeling the brunt of technological advancement, where human workers are being displaced by robots and automated processes that, like Project Debater, grow smarter and more potent by the week.
To them, Project Debater could elicit images of their very worst fears: the ushering in of an era, not only when A.I., machine learning and robotics surpass their own physical abilities, but also their intellectual capabilities. And while Project Debater is still a ways off from fulfilling futurist Ray Kurzweil’s claims that A.I. will reach human intelligence by 2029 and surpass human intelligence 16 years afterwards, there were moments on Monday evening when it became clear that Kurzweil’s predictions may well be on the conservative end. The Project Debater beat its human opponent in one of two debates on Monday — the second time it had done so since IBM began development in 2011 in Haifa, Israel.
IBM, which has seen its stock dip nearly 7% since the start of this year, has no immediate plans to market Project Debater in consumer applications, but the company suggested it could one day prove useful as a tool to augment human abilities — helping a lawyer prep for their closing arguments, for instance — rather than displace a human being entirely. That dovetails into many technologists’ arguments that artificial intelligence and associated technologies won’t outright replace humans but augment them or enable new kinds of jobs, just as manufacturing did during the industrial revolution in the 1800s.
As it stands, Project Debater is a startlingly impressive piece of technology and work-in-progress, not just for IBM but for the A.I. world as a whole. It will spark wonder — and some chuckles — but also, at least for some, worry.
More from JP