The most interesting point from the article wasn't about how well the AI's worked, rather it was the gap between peoples perception and their actual results.