This is a guest post by Jason Rothman. He has been posting a lot lately on my blog, and you guys seem to be happy with it, so here's another post!
When I was five, my parents took me to a fair where I played tic-tac-toe with a chicken and lost.
I got over it because... well, I was five, but I was reminded of that today when I read this article
about Nigel Leck who developed a robot, @AI_AGW
, that scans Twitter every five minutes for people tweeting negative things about global warming, and then answers them with scientific evidence showing them they are in fact wrong.
It looks for several hundred key phrases that climate change deniers use in their arguments, and then matches up the correct argument to their statement.
Basically, the premise is that "deniers" use pre-constructed phrases to argue against the mountain of evidence, and in doing so, they tend to use the same phrases over and over. Leck’s Robot finds those phrases and answers them with the scientific links to back it up. The machine learns.
Many of the people have no idea the argument is coming from a robot and continue to argue themselves into a corner. Sometimes the bot returns false positive results, because it cannot detect things like sarcasm, but it is learning ways around that.
It is just more fun when a robot does all the grunt work for you.
If you want to read more about machine learning, check out the wikipedia article for Turing Test