Meet Kia, the AI persona assisting an SFU professor's new course

Simon Fraser University’s newest class will have an unpaid expert on hand to answer questions about her field. And she’s only three months old.

“She” is anthropomorphizing the issue, though. Kia is an artificial intelligence persona developed by SFU’s iViz research lab, headed by professor Steve DiPaola.

And Kia, in what’s believed to be a world first, will be assisting DiPaola in teaching his new course: IAT 111: Artificial Intelligence Today & Tomorrow.

The class, which is open to any students, is an explainer of sorts. Its aim is to battle misinformation about AI and give students an idea of how to adapt it in their future fields.

“There’s so many issues going on that I really needed to be teaching this to a much more significant audience,” said DiPaola. “There’s such disinformation about what these systems are and what they aren’t. I think I gave over 40 talks last year to try and really explain to the general public what it is, what it isn’t, and how we can deal with it better for the future.

“This is the world that (my students) are moving into, and they really need to understand it. So by anthropomorphizing AI like this, I think people are just very comfortable to ask it questions. Its job is that we can stop and ask its opinion as an AI on an AI topic, and then we can even debate with it, which is kind of fun and exciting.”

He said Kia is not a replacement for any human role, nor will it be involved in grading papers or similar functions.

“Kia is there only to wait for us to ask her questions. She’s kind of an AI representative,” he said. “Imagine we’ve brought her in knowing that we want somebody who knows a lot about AI to answer questions.”

Kia will present as a young black woman with short-cropped hair, and is capable of real-time facial expressions and body language. She’s designed to be a sparring partner for DiPaola and his students as they explore the ethics and potential of the technology. She even has a personality — he describes her as honest and direct, and won’t veer into jokes or asides — and capable of injecting her words with the emotion she’s trying to impart.

She’ll either be projected through an overhead or on a large vertical monitor, depending on the classroom.

DiPaola has three aims for the course: Teach his students foundational knowledge, so they understand how AI works and what it’s capable of; show them what the future can look like, with Kia as an example; and demonstrate a level of practical use and how it may be translated into their career path in an ethical manner.

He’s also going to be running a similar program for faculty and staff.

“SFU wants to be an AI leader. We have the largest supercomputer centre in Canada that does AI,” he said. “There’s researchers like me and others doing all this AI, and we realized it can’t just be the AI researchers. It has to be the whole school. It has to be the teachers in philosophy and the staff. This is all about realizing that … everyone has to understand this stuff, and we all have to be around to vote and push our politicians on what we want and don’t want from AI.”

Classes for the upcoming semester and the next one are filling up quickly. Its popularity is clear, as DiPaola was at the tail end of a media-filled day on Wednesday when he spoke to Postmedia.

Pop culture is full of instances of AIs gone rogue — think Hal 9000, Skynet and the Terminator — that have created a negative impressions in much of the public, mirroring recent predictions from experts who foresee the programs killing off humanity once they become self aware.

Creators — whether musical, literary or artistic — lament the loss of their human content to web-scraping programs that then repurpose their work for pennies. DiPaola sees that much the same way DJs and song producers once viewed, as they sampled old works to create new art.

When it comes to the bias that AIs can present — Elon Musk’s chatbot Grok is a prime example — he points out that it’s the human-generated input that it draws from.

“I think, in a way, it’s nice to see that technology is poking right in our eyes, saying, ‘What are you going to do about it, Humanity?’ and we have to kind of really think through it.

“I think that the next 10 years are going to be quite significant in us thinking about these issues.

“ChatGPT and large language models, too many people think there’s some kind of oracles in there, that they’re all-knowing, when, in fact, it’s this amazing tool,” DiPaola added. “It’s just a very different tool. It’s a tool that can surf the history of everything that humans have written down.

“My standard line, especially about things like jobs and things like that, is, ‘You’re not going to lose your job to a AI. You’re going to lose your job to someone who knows how to how to control an AI or program an AI.’ … The tool isn’t going to take over. You just have to learn the tool.

“If we call AI a bogeyman for all these problems, and then we kill AI, the problems are still here. What I hope is the good side of the next 10 years is that we rethink this, maybe, there’s a way that we all somehow get money and three-day work weeks.”

jadams@postmedia.com