"AI should be used for improving human decision-making and not to replace it," says Moodley. "The danger lies in how we approach the technology. If we use it responsibly, almost as an early warning and recommendation system, it can add immense value. If, however, we rely on it fully to manage our health, then there are dangers. According to Rimmer, another factor is the scale of fallout that could occur. "If there is an error in the system, assuming it is a pervasive system, then everyone could get given the wrong diagnosis/information. One bad doctor can impact only a few people; one bad machine could impact thousands.