From medical research to school admissions, supply chains and insurance, fabricated data sets are lending a hand in interesting ways
Lina Colucci predicts a future where artificial intelligence-aided sensors in factories and warehouses spot hazards instantly, and where fitness buffs can bypass their neighbourhood gym in favour of an avatar workout coach who will count reps in real-time, correct lapses in form, and be available at all times.
As privacy laws have tightened in Canada – where a new federal privacy bill, with provisions for AI systems, waslast summer – and in other parts of the world, use of synthetic data is growing. By 2024, according to U.S. research firm Gartner Inc., synthetic data will account for 60 per cent of all information used to develop AI and analytics projects.
“The generated data doesn’t pertain to specific individuals but contains patterns,” Dr. El Emam says. “And when you analyze the synthetic data, you learn the same things you would have learned from the real data.” In Alberta, a not-for-profit organization called Health Cities has built synthetic data for a project aimed at preventing opioid addiction.
“Without this synthetic data, the VA’s ability to work with industry would be a bit handcuffed,” says Josh Rubel, chief commercial officer at Israel-based MDClone Ltd., which generated the synthetic data for the VA. “Now they’ve got 25 companies working with them on this project.”