Ng: Debating with a machine

WE HAVE seen that computers can handle almost infinite computations and memories.

The first large scale project to show that computers are intelligent was Deep Blue, where a computer beat then world champion Gary Kasparov in the game of chess. Then came Watson, in which the computer beat a human player in Jeopardy.

Then AlphaGo beat the world champion in the game of Go, an ancient Chinese board game that is said to have infinitely much more complex combinations than chess.

The latest attempt was by IBM called Project Debater.

Can machines not only understand human speech and craft answers to straightforward questions, but actually engage the person in an argument or debate the facts? Project Debater’s most decorated opponent was Harish Natarajan, the grand finalist in the 2016 World Debating Championships. They faced off during IBM’s Think Conference on Feb. 11, where Natarajan remained victorious. Compared to headline-grabbing wins such as IBM’s Deep Blue winning against a human chess champion, Project Debater failed to do the same. But it held its own.

Project Debater and Natarjan faced off in a traditional debate style where the topic was not announced ahead of time. Each side gave a four-minute introductory speech, a four-minute rebuttal, and a two-minute closing speech. This year, the topic was on the subsidization of pre-schools, with Project Debater arguing in favor of it.

After sifting through its data banks, it was able to keep pace with its human opponent and formulate arguments in direct response to the opposition’s points.

Although Project Debater lost, its presence on that stage also signified a victory. It was able to create an argument and break it down into cogent points, while backing it up with verified data. It was objective—perhaps a bit too objective—that you don’t forget that you are still arguing with a machine.

In today’s age of misinformation, technologies like Project Debater help us deal with the various nuances in our world. It is stepping away from the black-and-white world of artificial intelligence and into our own messy realities.