Can AI safeguard us against AI? One of its Canadian pioneers thinks so