Dreaming of basic AI, but scared of advanced AI --- simple AI to reduce medical errors, or to make medical workers more efficient, exists already. I personally trust a machine more than my own or a colleagues' tired eyes to detect the small node on a X-ray or the little hook on the otherwise normal ECG, and then use my brain to make sense (or not) out of it. But these simple tools still have a difficult life in our daily life. I think it's because it's tempting to rely only on it and to switch off our brains (I have seen that...), or/and because it is feared that this will lead to a reduction of the workforce, without acknowledging that the remaining task, namely the synthesis, is more demanding, so a balancing calculation should be done (which is not the strength of some health directors). But having seen what already basic AI can do as harm -as for example the panic of a patient whose watch told him he might have a heart attack, rising his heart rate so much because of stress that he augmented his risk for that- I'm a bit worried of what advanced AI can do without proper preparation.

Source author: Katharina Beyer katharina.beyer@etu.unige.ch