Caveat: debate-o-matic

One of the subjects that I taught to my students in Korea that I considered most valuable, both for the English skills it engendered as well as for general thinking ability, was debate.
I was the “debate teacher,” and I was well-known for even turning lessons otherwise structured into impromptu debates. The kids mostly seemed to get something out of it.
So now… they’re trying to make an AI (artificial intelligence) that can do debate – in the same way that we have machines now that play chess or baduk (“go”), that diagnose medical conditions or explore other planets. This is just another small step.
I watched this video.

I am both disappointed and impressed. This is often the case when confronting these odd black boxes that computer engineers are constructing these days. They can seem preternaturally smart and eerily stupid at the same time. The AI participating in this debate clearly had a lot of facts to hand, and was reasonably competent at marshaling them in a well-structured argument. But it missed the key thrust of its human opponent’s argument, and thus its rebuttal almost failed to make sense. I was somewhat annoyed that the moderators, who spent time afterward discussing what they’d just done, failed to bring this up.
picture

Back to Top