You are viewing a single comment's thread from:
RE: Curating the Internet: Science and technology digest for November 22, 2019
AI is gonna be key to many tech developments, and particularly in decentralization. But I find psychiactric diagnosis by AI not only utterly ridiculous, but Orwellianly terrifying. I don't even think real human psychiatists are better than guessing at what consciousness is and how it goes wrong, and now their handing off their guesses to AI? Just let me check off boxes on a listicle.
Thanks!
Thanks for the reply!
I'd feel much better about psychiatry if they could eventually get to the level of identifying and treating bio-mechanical root causes, as I suspect that a number of totally unrelated physiological problems can cause similar seeming behavioral anomalies. I could definitely see AI helping with identifying those underlying root causes, although that's not what the article is about.
As it stands now, my observation of psychiatry over the years is that - in practice - the doctors seem to try medicines until something works, and as they do, they change diagnoses to match the medicine they're trying. Then, whatever medicine eventually seems to cause improvement tells them their ultimate diagnosis. So I don't worry about them adding an AI to the diagnostic toolbox. To me, it doesn't seem like the process can get much less accurate. I don't think an AI should be the only diagnostic decision-maker, but I am ok with it being another input for consideration by a human doctor. Of course a risk with that is that doctors may come to trust the AI's too much, as we have seen with accidents that occurred during the emergence of autonomous vehicles, when their human observers were less than attentive.