South Korean researchers taught AI to admit when it doesn't know something, mimicking how human brains develop.
This breakthrough could make chatbots safer for medicine and self-driving cars.
Want to know more?