But it sounds like with AI doctors did better overall, or that is how I read the first couple of lines. If that is true, I don't really see a problem here. Compilers have eroded my ability to write assembly, that is true. If compilers went away, I would get back up to speed in a few weeks.
The compiler analogy is seductive but problematic imo.
A compiler can be fixed thing that does a fixed task. A cancer recognizer is something like a snapshot of people's image-recognition process during a period of time. These are judgement that can't be turned into set algorithms directly.
There was a discussion a while about how face recognition trained with Internet images has trouble with security camera footage 'cause the security camera doesn't certain images.
It sounds weird to say that what cancer looks like drifts over time but I'm pretty sure it's actually true. Demographics change, the genes of even a stable group change over the generations, exactly how a nurse centers bodies, etc. change over time and all these changes can add to the AI judgement snapshot being out of date after some period. If the doctors whose judgements created the snapshots no longer have the original (subtle) skill then you have a problem (unlike a compiler whose literal operations remain constant and where updating involves fairly certain judgements).
Doctors’ ability to taste diabetes in urine has also probably eroded since more effective methods have come on the market. If they’re more accurate with the use of AI, why would you continue without it?
Every time I see "but your skills will atrophy" arguments like this, they always leave an implied "and you'll need them!" lingering, which is a neat trick because then you never need to explain.
However, I would like someone to explain this to me: If I haven't needed these skills in enough time for then to atrophy, what catastrophic event has suddenly happened that means I now urgently need them?
This just sounds very much like the old "we've forgotten how to shoe our own horses!" argument to me, and exactly as relevant.
I know nothing about this field, and the actual paper is behind a paywall, but it says that after the "exposure to AI", the adenoma detection rate (ADR) dropped from 28.4% to 22.4%.
As a layman, does ADR simply mean suspicion, or does it mean they correctly and accurately saw adenomas in 28.4% of patients before and now the rate is only 22.4%. Or just that they suspected it 6% more before? Does the actual paper detail if they simply stopped seeing illusions, or did they actually stop seeing meaningful things they used to see?
I'm sure the paper goes into more detail, but I'm more interested in the false positive vs false negatives than just overall %.
Misleading title. Eroded ability to find precancerous colon polyps, not cancer. I am a gastroenterologist that consults at the local VA hospital. They started using Medtronic GI Genius AI last year across the USA. It has a very high false positive rate for true polyps so I suspect this is playing a role in this study in addition to doctors becoming overly reliant on it. Overall I think it is a promising technology once they refine the models.
12 comments
[ 4.1 ms ] story [ 44.0 ms ] threadA compiler can be fixed thing that does a fixed task. A cancer recognizer is something like a snapshot of people's image-recognition process during a period of time. These are judgement that can't be turned into set algorithms directly.
There was a discussion a while about how face recognition trained with Internet images has trouble with security camera footage 'cause the security camera doesn't certain images.
It sounds weird to say that what cancer looks like drifts over time but I'm pretty sure it's actually true. Demographics change, the genes of even a stable group change over the generations, exactly how a nurse centers bodies, etc. change over time and all these changes can add to the AI judgement snapshot being out of date after some period. If the doctors whose judgements created the snapshots no longer have the original (subtle) skill then you have a problem (unlike a compiler whose literal operations remain constant and where updating involves fairly certain judgements).
My solution is increase the amount I write purely by hand.
I’m sure similar things have been said with:
- calculators & impact on math skills
- sewing machines & people’s stitching skills
- power tools & impacts on craftsmanship.
And for all of the above, there’s both pros and cons that result.
However, I would like someone to explain this to me: If I haven't needed these skills in enough time for then to atrophy, what catastrophic event has suddenly happened that means I now urgently need them?
This just sounds very much like the old "we've forgotten how to shoe our own horses!" argument to me, and exactly as relevant.
https://news.ycombinator.com/item?id=44883350
https://www.thelancet.com/journals/langas/article/PIIS2468-1...
https://doi.org/10.1016/S2468-1253(25)00133-5
If someone finds a link to a pre-print or other open access, please post it in the thread, as this is just the abstract.
As a layman, does ADR simply mean suspicion, or does it mean they correctly and accurately saw adenomas in 28.4% of patients before and now the rate is only 22.4%. Or just that they suspected it 6% more before? Does the actual paper detail if they simply stopped seeing illusions, or did they actually stop seeing meaningful things they used to see?
I'm sure the paper goes into more detail, but I'm more interested in the false positive vs false negatives than just overall %.