The Artificial Intelligence Conference in Boston this week was, well, a trip. I learned a lot and I think I produced two really good stories out of the ordeal. I also took a lot of pics that I placed in my blog photo albums.
As for analysis. First let’s get this out of the way right now. People in the AI field are smart. I mean, freak’n smart. I think this is due to vast and various fields of academia that AI relies on — computer science, statistics, robotics etc. It’s official, I don’t think anybody is smart until they build their own semantic web search — one of the many many many different AI applications I discovered at the conference.
I can barely report on everything I saw, so I’ll just hit the high points. Or, if you are a visual learner, just check out the photos and smile.
The poker tournament.
Yes, a poker tournament of robots. The idea here is that chess has been mastered by robots. But poker is a different beast all together. In poker there is bluffing, and there are variables that are unknown and random — what does your opponent have and what cards are next. In chess you already know all the moves possible for yourself and your opponent.
After talking with one of the students who created a poker-bot, I was told that yes, robots can bluff and can call people on their bluffs. Although it’s not a psychological bluff the way we normally think of bluffing — its cold and calculated. If you have a really good hand, you raise. At the same time, there is a threshold, where you can have a bad hand and a certain amount of chips in the pot (pot committed) and you decide to bluff.
Me personally, I tend to bluff just when I’m feeling snarky, but I guess that’s why these bots would probably beat me.
Robots on Parade
The rest of the physical robots were cool to check out. It’s funny, I expected really advanced stuff — as if Data from Star Trek the Next Generation was going to be there. Or at least Johnny #5 from Short Circuit. Silly me. The truth is making robots that can see, hear and understand things is tough. Making robots that can move gracefully is also hard. Almost making any type of robot that thinks for itself is just still too hard to do connivingly.
The problem lies in the types of knowledge we take for granted all the time. When I walk somewhere and talk to someone, I know that gravity will keep me down, that two things can’t occupy the same space at the same time, and that the person won’t hear me if I start typing my words. These are commonsense things that children know, but what I’ve learned repeatedly at this convention (mentioned in both stories at one point or another) is that it’s these basic commonsense bits of knowledge that are hard to teach robots.
They can learn everything a Ph.D. should know, but they can’t grasp the important aspects of life a child understands. Perhaps that knowledge set is too ingrained in our conscious and experience to program as of yet.
Musical Bots
The musical AI programs were also a trip (read the second story for more complete explanations). One AI program would make melodies that sounded human. Another produced musical accompaniment that matched the changing melody. In other words by taking past data and current circumstances these robots would predict what sounds to make in the future, so as to make beautiful music. We in the human world call that jamming.
I really wanted to get the two AI programs to jam together, producing the first robot meta music session (and making my first millions), but it seemed out of ability to organize, so I didn’t try.
Me personally, As if you cared
As for me. This was a fun experience. I got to see Boston. Well, I got to see a convention center in Boston. But I also got to try my hand at some quick reporting. On Thursday it will be one week since I got the green light on this story. I produced two 1,000 word stories in one work week. But the best way to give you a sense of timing is this…
I did all the info gathering for one story on Thursday (I was lucky to get such good sources in one day), wrote the story for Friday morning edits and it was up on Monday. On Monday after class I headed to Boston. Tuesday I did reporting for the second story, Wednesday I did writing and more reporting and here on Thursday the story is up.
While there I also met a CNET writer and a New Scientist writer. Both are really good news organizations. I expect to see good coverage from them. I hope it’s not better than mine, but what can you do. I will tell you one thing. If all the stories on this convention are as bad as this AP story, I’ll be in good shape — man is that a bad story.
For now I’m going back home to Brooklyn. I might have another assignment, but we will see. Soon I’ll be traveling back to Los Angeles (home of the breakfast taco) to see friends and family. Posting might be light until early August — since I’ll be in the woods for about a week. But I’ll be in touch.
In the meantime. Just know you shouldn’t talk trash about the robots. They are keeping score.