Canada’s most valuable assets include our society’s values, respect for others, equality and diversity – and those are also big issues in artificial intelligence (AI) says Simon Fraser University (SFU) computer scientist and AI4ALL director Angelica Lim, who is leveraging the power of data in AI to ensure a diversity of perspectives.
As a professor of professional practice in SFU’s School of Computing Science, Dr. Lim stresses the need for diversity of people working in AI to counter any potential bias. She notes that while nearly 40 per cent of computer programmers in the 1980s were women, that number dropped significantly, while the participation of women in other scientific fields continued to rise.
According to the Information and Communications Technology Council, the tech industry is comprised of 25 per cent women and 75 per cent men – despite the Canadian workforce being made up almost equally of men and women.
SFU is the only Canadian university participating in the AI4ALL initiative, aimed at closing the gender gap and supporting educational programs designed to encourage diversity. Other partners are Stanford University, Princeton University, the University of California Berkeley, Boston University and Carnegie Mellon University.
Diversity has a direct impact on Dr. Lim’s own research into the personalization of robots.
“Big data takes aggregate information from a large number of people,” she says, adding data for her work is emotion data.
“For example, I study many people showing happy emotions or other kinds of emotions that we want to detect,” she says, noting there are differences in how children express emotions compared to adults, and differences in Eastern and Western cultures.
“[However,] it’s not just what’s common between people that’s important, but also our individual differences. This is something that we’re looking into – the diversity aspect. As diversity is one of Canada’s strengths, it is important for me to not simply wave a giant brush over everyone and take aggregate statistics, but also look at pockets of differences between groups and people,” she says.
In addition to her work as a roboticist, Dr. Lim focuses on affective computing – affect being emotions, moods and feelings – and developing robots or devices that can recognize these affects.
This goes back to Dr. Lim’s work in Japan and France, developing software and cognition models for robots to interact with humans. She is exploring this through neuroscience, machine learning and developmental psychology, and has built software for several robots including Softbank Robotics’ Pepper the humanoid robot.
But while Dr. Lim continues her research into robots that can interact seamlessly with humans to help them, she is also working alongside projects at SFU’s Big Data Hub, including one trying to revitalize the Blackfoot language through interactive chatbot technology.
“The hope is that if we’re able to digitize some versions of the language, then it can continue to live and be taught to future generations through technology. It would require big data in terms of general language models, but also small data to specialize the model for Blackfoot.”
Research is often 20 to 30 years ahead of when it reaches the market, says Dr. Lim, adding it takes time to distill the complexities of robotics into an application that people can use every day.
“Research that was done in robotics in the 1980s is slowly coming to autonomous cars, and the AI technology for speech and natural language processing that began over 30 years ago is now used in products like Alexa,” she says.
“That is what’s great about research: you’re able to explore technologies without a business-driven intent. Researchers are not there to make money; they are there to progress science, and hopefully that technology will be groundbreaking, to produce an innovation we had not anticipated.”
Produced by Randall Anthony Communications. The Globe’s Editorial Department was not involved in its creation.