There is a place on North Campus where rabbits spend their days hopping around and eating grass. Some rabbits - the Peter Rabbit, take-charge types - are leaders. When they move to a patch of grass, their brethren follow, eat until their little fuzzy bellies are full and survive to hop another day.
One day, the leader rabbits didn't show up. The other rabbits twitched their noses and pulled on their ears and, without their leaders, went out of their way to avoid the grass patches. In fact, if they even got close to a patch, they would start to tremble.
Eventually, the shy rabbits wasted away from hunger. The end.
Did I warn you this was a sad tale? Don't worry, it really isn't. These rabbits don't live under a tree on Quad; they live in a computing science lab. More specifically, inside a computer in a computing science lab, where the leader rabbits are orange triangles, the followers are purple triangles and none of them actually has a fuzzy little tail to twitch. The fact that there are even leaders and followers at all is a surprising twist in an artificial intelligence research project led by Vadim Bulitko, a computing science professor.
The project looks for surprises and researchers certainly got one. All the rabbits are programmed to automatically start eating when they come across a patch of grass. The researchers didn't expect leaders and followers to emerge; it just happened. "Who knows what's going to happen next?" asks Bulitko. Who knows, indeed? In future stages of research, the team hopes to introduce "utterances," which the rabbits and wolves could eventually transform into a language.
You certainly know by now that AI is a big part of life in the 21st century, and you might have heard that the U of A's computing science department is among the top in the world when it comes to AI research, especially the branch called machine learning. For the Spring 2018 issue of New Trail, our team had the chance to dive into the topic and bring our questions to some of the field's greatest minds. Questions like:
- How could AI change us?
- How close are we to the singularity, anyway? (Really, really far away.)
- How do we keep AI from running amok and breaking humans down to their raw atoms to create more paper clips? (A stop button is the simple answer, according to Mike Bowling, professor of computing science and AI card shark.)
- And, of course, we answer that burning question: Who is closer to reality, R2-D2 or Hal from 2001? 2001? Can you say, "OK Google, open the pod bay doors?"
Being part of the U of A community gave us an excuse to poke our noses into some really cool, cutting-edge research, some of which you can read about in the current issue, available on newsstands in SUB. One thing we heard again and again was that AI research is a long way from our pop culture conceptions - at least on the outside.
"Lower your expectations," warned Spencer Murray, communicator for the U of A's Alberta Machine Intelligence Institute. He was speaking just before giving New Trail art director Marcey Andrews and managing editor Karen Sherlock a quick tour of an AMII lab, but it was a thought repeated throughout our research for this issue. Even though AI research at the U of A is at the forefront of the field, don't expect to find C-3P0 greeting you at the door. The real magic happens in the programming, which the rest of us can't see. And that, if you think about it, says a lot about how far most of us are from grasping what AI is and what it can do - not to mention what it can't docan't do.
Lisa Cook - Associate Director of Communications, Office of Advancement
Lisa is the Associate Director of Communications with the Office of Advancement Communications. She serves as Editor-in-Chief of New Trail, and is curious to see who spotted the one robot that's not like the others on the cover of the Spring 2018 issue . Let her know!