Using Human Ethics to Power Smarter Machines
Technology Montreal firm believes Artificial Intelligence has the potential to do plenty of good for Canadians.
AI was conceptualized in the 1950s, but over the last decade, major advances have allowed for implementation in business and consumer-facing technologies. AI already powers several applications that we use daily and the transformative technology can bring sizeable benefits to businesses and economies by accelerating growth, enhancing productivity, and driving innovation.
AI and ethics
When discussing AI, the topic of ethics often arises. Stradigi AI, a leading AI firm based in Montreal, believes that AI solutions built with an ethical framework will make a lasting, positive impact on Canadians, and is at the forefront of the industry by being committed to building ethical AI platforms. “Attaching ethical mandates to machine learning can make a significant impact on how businesses and people make decisions,” explains Carolina Bessega, Chief Scientific Officer at Stradigi AI.
AI uses massive, existing data that the machine interprets and learns from, delivering conclusions based on what it identified. Since existing data is used, a subset of Ethical AI is the recognition, and elimination of bias. As smarter machines begin to enhance human decision-making, Bessega and Stradigi AI build ethical AI platforms that employ algorithmic fairness, a theory looking at the consequences of algorithms that make unfair decisions because they lack instinctive, human nuances. “Having a strong understanding of potential biases is critical to the development, implementation, and monitoring of AI technologies,” says Bessega. “Even though initial data training may even the field at the start, it’s imperative to monitor the algorithm as it may recognize a bias that was not at first identified.” It’s important to understand that the purpose of AI is not to replace human-decision making, but rather to augment it. Bessega believes that, without a regulatory framework yet within the field, ownership and accountability needs to be embedded into the end-to-end process, shared between the researchers creating the AI, and the business stakeholders implementing it. “I’m sure a company that develops a program without diverse data is not doing it on purpose, but they could still deliver a biased product that renders it unethical because it didn’t cover all scenarios,” says Bessega.
Lending AI credibility
Stradigi AI builds custom AI applications for enterprise clients, but to add credibility to the field and do good with AI, they’ve worked on two projects inspired by the UN’s AI for Good initiative, a platform that fosters dialogue around how AI can be used to solve some of society’s biggest challenges.
The first, produced in partnership with the Deaf Anglo Literary Centre in Montreal, uses AI and computer vision to help people learn American Sign Language, and better communicate with the hearing-impaired. The second, developed with the Centre Hospitalier de l’Université de Montréal, uses AI to predict epileptic seizures.
“What if, as an epileptic patient, you could know in advance that you’re at risk for seizure?” Bessega asks. “You could alleviate the stress of the unknown, and be prepared to ensure you’re safe, or avoid the seizure itself. There are so many possibilities with AI.”
Bessega is hopeful AI will help advance medicine. By working collaboratively with health care professionals, AI could help in prevention, earlier diagnostics and treatment. She believes AI could contribute to the development of new medicines and help those living with mental health issues.
“AI gives us the freedom to be innovative and creative, and to see the world in different ways,” she says. “We have to make these algorithms credible and fair while ensuring that we educate people so they’re not scared. We’ll continue to do even more research to make ethical AI work, because we believe it can positively impact the world.”