Balancing Optimism and Skepticism in AI Medicine

I feel like you and your work chart this path between being

hopeful about where we're going and being skeptical about where we

are now.

How do you apply your kind of ability to balance optimism with

skepticism when it comes to AI in medicine?

Yeah, I mean, my main theme there is I wanna see absolute

validation. I wanna see really hard evidence.

So what's disconcerting is we do have that, like for

example, AI mammography, you know, over

100,000 women in Sweden, uh, with an algorithm that's available here in

the US, and that should be part of every

mammogram.

The same holds true for colonoscopy.

Why would you want to have a colonoscopy when you have 44 randomized

trials that show that this picks up adenomatous polyps by

having machine vision during the procedure?

So what frustrates me is that where we have

solid data, it can't get any better than that,

we're not implementing,

and then we are implementing, you know, these chatbots, uh, widely

without the evidence. So this is a real paradox.