Wednesday, November 26, 2008

Open the pod bay doors, Hal. I'm afraid I can't do that, Dave

From the Chronicle: Machines Become Teachers

Computers will become better at teaching than most human professors are once artificial intelligence exceeds the abilities of people, argues Ben Goertzel, director of research at the Singularity Institute for Artificial Intelligence, in Palo Alto, Cal., a private organization promoting Mr. Kurzweil's ideas.

These new computer teachers will have more patience than any human lecturer, and they will be able to offer every student individual attention — which sure beats a 500-person lecture course.

Sure, one-on-one human teaching will always exceed a computer-student experience, Mr. Goertzel acknowledges, but what college undergraduate gets a personal tutor these days?

(In answer to the last question: maybe students who go to the Writing Center or Math Center on campus?)

The article has some ideas that don't seem applicable to an MLA discipline (virtual research assistants?) but that may be useful in the sciences. You'll be shocked to learn, too, that the Internet has made research easier and that Google is really, really awesome.

On the other hand, the "computers will replace teachers" argument given here, like a lot of the many, many articles written about this since the early 1980s, relies on the either/or fallacy: either an impersonal 500-person lecture hall or a personal, albeit virtual, tutor. There's no mention of a discussion-based class, no mention of the human interaction that takes a class to unexpected places and makes it memorable.

I think the real question here is what is meant by "teaching." A long time ago, the idea was "computers will replace teachers because we can sit students in front of terminals and make them practice verb tense endings until their eyes bleed." Drill and kill, it was called. That was individual attention, but not in a good way. Does "teaching" mean having the infinite patience to impart a piece of information until the student gets it? That seems to be the model being proposed in the article.

Computers have already transformed the way we teach (no duh), but more for their communications functions than anything else. I think the real transformation will probably be more like a "questing together" model, something like World of Warcraft or some other game seems to be. I don't play it but have read about it, and it seems to me to be closer to what happens in a class than the "patient tutor drilling students" model.

So who's with me for a quest to find Hester's missing A? We'll have to get by Chillingworth first, but if we group together, we can cast some spells, knock him out with his own potions, and make it in time to save Dimmesdale.

4 comments:

Bardiac said...

Interesting!

Except having gamed a bit, I think learning happens in ways that might surprise non-gamers. There are folks who pick up the game as you'd expect, a sort of trial and error, applying what one learns and overcoming frustration. Lots of folks start out gaming that way, relatively few stay there.

In the games I've played, a LOT of information gets passed between avatars/players. I started out doing trial and error, then someone would explain something, answer some questions; people become limited "experts" at the stage where they are. I think most of the people I knew did a lot of learning that way.

But, the surprising learning for me was that the more expert gamers spent a LOT of time reading up on the game, especially on mechanics, and then figuring out in all sorts of ways (including writing programs, even programs that added into the game to provide information and data collection). So they were actually learning in really traditional ways, reading up, often about very complex stuff, and then applying it, making it more accessible to others (especially in their guilds). I knew people who spent hours off line reading and figuring out the math of encounters and such. Amazing.

The thing is, the rewards are pretty quick and obvious in games, and they aren't in real life. No one gets a brand new shiny coat after passing a quiz that says to everyone who looks: I passed Quiz A!!!!!!

undine said...

Bardiac, that's interesting. The kind of knowledge model you're describing--those who know more teaching the newbies, those who are experts WANTING to know more and reading about it--is exactly the kind of communication that would make for a really exciting teaching model. Once I learned (via a family member) that WoW players can talk to each other online as they play, this seemed to me to have great possibilities.

I would love to have an environment where they could get a shiny new coat and increased powers for passing Quiz A!

Dance said...

Well, seriously--what *isn't* going to be obsolete once artificial intelligence is able to outthink a human brain? When AI gets to that point, I will happily be replaced by a computer. Believe me, I am trying as hard as I can to articulate the processes by which students can teach themselves and not need me. But when AI gets to that point, we won't be able to tell computers from humans anyhow.

Happy Thanksgiving!

undine said...

Happy Thanksgiving, Dance! I still wonder if AI will be able to keep up with the really unpredictable kinds of thought leaps that students make, especially those that occur because of cultural shifts. For example, a few years back students started asking me if Homer, in "A Rose for Emily," was gay. It's pretty clear from the context that he's called a "man's man" and is something of a womanizer, which would be in keeping with the period of the story and with Faulkner's portrayal of him, but familiarity with contemporary gay culture had given the students an entirely different and really interesting way to think about the story. Those kinds of cultural shifts would be difficult for a computer to absorb, I'd think.