When I went to meet Ryan Small at his home in Auburn, I expected to wade into a number of stereotypes. Matrix posters on the wall. A landscape of computer parts scattered throughout a cluttered apartment. A few hundred pizza boxes marking the times he remembered to eat in the midst of all that geeking.
Instead, I found him in a condo that was nearly immaculate. A chocolate Lab and a vociferous cat greeted me at the door. Ryan’s girlfriend was lounging in a room upstairs.
“Are they real?” I asked him referring to the pets. “Or things you created for company?”
A stupid joke that fell flat. But you have to know what this 24-year-old University of Southern Maine student is about in order to understand my attempt at humor.
The facts are these: In the middle of May, this young Auburn man won an artificial intelligence competition at the 2009 World Congress of Evolutionary Computation in Trondheim, Norway.
Evolution. Computation. Artificial intelligence. When I was 24, I was pumping gas.

On the surface, Ryan’s work doesn’t sound too extraordinary in today’s highly computerized world: At the big showdown, he introduced a program to play a first-person shooter game called “Unreal Tournament.” But enter Agent Smith, the revolutionary, fast-learning protagonist of the game.

The two of them didn’t just prevail at the big show. They blew the crowd away.
“Ryan gave his talk to a standing-room-only group of about 70 people and received nonstop questions of interest afterward,” said his professor, Clare Bates Congdon. “It’s nearly unheard of for undergrads to present at international research conferences. And he did very well with the whole thing.”
You want more macho to pump up the wow factor? Ryan’s AI at the big show won it all by beating out another by the name of Agent Chuck Norris.
So take that.
Ryan didn’t simply design a computer game character. This 24-year-old, who learned computer language at about the same time he began talking, created a character with a form of intelligence that can adapt, learn and evolve.
Ryan attempts to explain this artificial intelligence thing to a lay person with great patience.

On his computer, he pulls up a long list of what appear to be random numbers and letters. If you took the alphabet and a few numerals and flung them at a sticky piece of paper, this is what it would look like.
“This is how Smith sees the world,” Ryan says. “This determines how he is going to behave.”

To Ryan and Agent Smith, this is a collection of different behaviors called rule sets. Each type of behavior dictates how Agent Smith’s AI is going to react in various
situations while doing battle. If his health is good, he might go for broke and
fire away, for instance. With another set of behaviors, he might choose to do the opposite.

Whatever the decision, it happens at something near light speed. Some of the decisions will be
good and will return good results. Some of them, not so much. Either way, Smith
will learn from his results, based on Ryan’s programming, and for that reason make different decisions later.

To look at the rule set on a computer screen is a little daunting. It’s like
trying to read the mind of this artificial being and then describe what you find
there. But I see it this way: Every one of us in the real world has made good
decisions and bad ones. We have modified our behavior based on the consequences
of the things we have done and continue to do so
all the time. We have learned from
our mistakes and we try to pass those
lessons on to our children.

It’s not so different for Smith, really, aside
from the apparatus of reproduction.

The intelligence Smith enjoys thanks to Ryan’s work can be improved upon. The
best traits of the rule sets can be recombined and passed along. If Agent Smith starts
popping out AI children, in other words, they will likely be smarter than their old
man.

“The system is reproducing a new child (solutions/rule sets),” Ryan says, “and hopefully
making it better.”

Like natural selection initiated by a grand creator. If you wanted to use the concept to spur religious debate, it would not be tough to get there.
But let’s not.
The science behind his work is impressive – you don’t find lightweights of science at the 2009 World Congress on Evolutionary Computation, after all.
But equally fascinating is Ryan’s approach to his work. Ten seconds into a conversation with him, it becomes clear that this is much more than gaming. If it was inconsequential child’s play, Ryan might not be interested.
“I’m awful at the games, myself,” he says. “Especially the first-person shooter games. But games are an easy entry point into the world of artificial intelligence.”
What he is good at and what he loves (his eyes literally blaze with something like ecstasy when he talks about it) is programming. And programming an intelligent system that can evolve, learn and benefit from mutations is far from child’s play.
“Mutations help keep diversity in the population,” Ryan says. “Mutations help it to find global optimum.”
That line was so far over my head, I didn’t even hear it whiz by. For a simpler explanation of what Ryan’s work means in a larger sense, I turned back to his professor.
“The design of these artificial intelligence programs,” says Congdon, “has practical applications in that these systems are seeing increasing use ranging from military applications, to Mars explorers and entertainment.”
Ryan and his manufactured intelligence could someday influence what happens on battlefields or on far-flung planets. Yet there is no sense of loftiness about him at all. When asked to name his scientific hero, it is not Einstein or Feynman or Schrödinger, whose name he utters. It is Congdon, and it does not smack of sycophancy.
“If it had not been for her,” he says, “I would never have gotten involved in artificial intelligence. She’s been a great push. At first, I planned on keeping with it more as a toy, but she kept pushing.”
Something else surprised me about this blooming genius (he laughs at that word but I’m comfortable using it) as well. Bucking the tradition of geeks everywhere, the fondness he feels for his professor, his girlfriend and his pets doesn’t extend to the artificial beings that occupy his mind.
You would think that Ryan would love his creations a little bit. Agent Smith, after all, has garnered him worldwide recognition by kicking butt all over Europe.
“I have no real personal attachment to Smith,” he says. “I don’t look at him so much as a creation. It’s all about the intelligence behind it.”
Intelligence, yeah. He jokes that the first time he was asked as a child to spell his own name, he muttered “R-Y-A-N enter.” His dad got him into computers as a youngster and he thought everything needed to be followed by a poke of an enter button.
Brilliant folk always have such great stories.
As a journalist, my most pressing thought is: What will this brilliant young man do with these enormous talents? Will he be lured by a government think tank and spend his days creating masterful algorithms for the good of the nation? Will he design the robots that will someday take care of all our heavy lifting?

“I enjoy programming to make a career out of it,” Ryan says. “But honestly, I stay away from the robots. I’m not an engineer.”

Ryan seems to give this great thought. Remember the Hollywood genius who did all that big work for Skynet? Linda Hamilton shot him in the hand and then the poor dude had to spend the rest of his day with an Austrian actor with a terrible accent.

Yet Ryan’s ambitions are no less lofty. Most of us understand that for mankind to thrust off into the future, it will require great minds standing on the shoulders of scientific giants before them. If Ryan’s skills were resigned to the development of home entertainment alone, it would be a loss to the rest of us. There are agencies out there tasked almost exclusively with making sure we advance as a species and they need more than anything, young and gifted minds.

“I’ve flirted with the idea of working with NASA,” Ryan says, “or possibly DARPA.”

DARPA – the Defense Advanced Research Agency – is considered the technological engine behind the Department of Defense. The agency designs projects that most of us won’t hear about until a decade after they are put into use. 

With still roughly two years left at USM, Ryan seems a step closer to NASA. Congdon, his professor and hero, has contacts at the space agency which this very day is pondering ways to safely transport men, women and probably some chimps to Mars. Most of that technology has not yet been perfected or even designed. Ryan might be one of the people who helps create the next great life-support system, for example, that makes it happen. And possibly sooner than later.

“I’ve been given the opportunity to play around with some of the NASA software,” Ryan says.

Most 24-year-olds I know tend to play around with 24-year-old girls or possibly a Camaro.

I have photos of Albert Einstein hanging in my room, you know. Like most of us with an average IQ, I find the concept of high intellect both dazzling and frightening, a force of nature that can send our species off into a bright future or bring the sky crashing down.
Ryan Small doesn’t have the wild hair or plans to build bombs. Like Einstein, it was curiosity that led him to the science. Albert wondered what would happen if you switched on the headlights while traveling at light speed. Ryan marvels over the splendor of natural selection. Darwin, perhaps, meets Bill Gates.
“I’ve always found the concept of evolution quite elegant,” he says. “I’ve always loved problem solving.”

Copy the Story Link