When nearly perfect AI agents work alongside more idiosyncratic humans, there’s a chance for things to go wrong.

Karim Hamade, a 2024 MSc graduate in computer science, went in search of a solution.

As a member of the Computational Social Science Lab where he was advised by Assistant Professor Ashton Anderson, Hamade proposed a formal evaluation framework to assess the compatibility between AI agents and humans in these increasingly common interactions.

Using chess as an example, he looked at how AI can be a partner to humans in selecting moves, for example, suggesting moves or pieces to lower-skilled players and seeing how successfully they can follow up on those suggestions. He also looked at how AI can act as a coach by playing against humans in a specific way that forces them to address their weaknesses.

“I’ve long been fascinated by how AI in games like chess makes so little sense to us despite being strong, and so much potential collaborative use of AI in complex environments is held back by the fact that we often can’t follow up on the machines’ decisions,” says Hamade.

Hamade also did a summer internship in finance where he designed an algorithm to detect and rapidly capitalize on short-lived trading opportunities in the market. After graduation, he will join financial firm IMC to work as a quantitative researcher.

He hopes his quant job will fulfil his goal of working “in a fast-paced, competitive high-stakes environment” and longer-term, he sees himself working on AI.

“There is so much to be done in terms of making it work for everybody,” he says.

Hamade looks back fondly on his time at U of T, with particular appreciation for his advisor, Ashton Anderson, and all the professors he has encountered.

“The courses are cutting edge without being overly dense that they interfere with research, and all the staff I interacted with were helpful and motivated.”

His advice for incoming master’s students is to “explore as much as possible.”

“You design your experience, so take courses in areas that interest you even if they’re unrelated to your research, and I strongly suggest taking time off to do an internship to get a taste of applied research,” says Hamade. “Be proactive with your research, you have the exact autonomy that a PhD student has, which is really unique.”