IBM recently unveiled a new artificial intelligence system that can craft lengthy, persuasive arguments during a debate.
The company unveiled the system, called “Project Debater – Speech by Crowd,” during the tech showcase CES earlier this week. At the show, it presented arguments both for and against making gambling illegal. It later did the same for the idea of developing autonomous cars.
Though it can’t come up with its own opinions on topics like gambling or social media, the system could become an important step along the way to AI that thinks and communicates how we do.
Under the Hood
Project Debater does little writing of its own — the AI system’s core strength comes from categorizing and analyzing the arguments fed into it by humans.
For instance, Project Debater would recognize that the statement “Social media enables us to keep in touch with long lost friends and family” as an argument that social media is good that pertains to the topic of communication.
The algorithm scans short human-written arguments on a potential debate topic for specific keywords and language, according to a YouTube video published by IBM. Then it groups those arguments by topic, determines whether they’re for or against the proposal, and organizes a speech based on the arguments it analyzed.
That’s not to say that IBM built a computer that has thoughts and opinions of its own — the tool is more of a compiler and language processor. But that doesn’t make it any less impressive.
Language is notoriously difficult for AI to grapple with. Sure, any internet-connected device can run words against a dictionary, but a phrase’s context, nuance, and sarcasm constantly sail over even our best machines’ heads.
And if IBM’s project can produce coherent arguments, it’s already ahead of practically everyone on Twitter.
READ MORE: Behind the Scenes of Project Debater – Speech by Crowd at CES: Part One [IBM Newsroom]