Kind of feels like the same crowd that convinced itself that token-maxing was the right way to use AI is now trying to convince everyone that scan-maxing is healthcare.
This debate revealed a lot about how information is handled by the medical system (LOL, the line about medical doctors and the medical establishment). For this particular technology, it may be right that it will not lead to better outcomes, but I contend that there is a lot of good information that is not obtained because of the way that information gets co-opted by the medical-industrial-legal complex, creating testing cascades, etc. To correct this, healthcare needs to move beyond Bayes rule toward decision theory. The other parties - industry, legal, etc - need to change as well.
I've got a more optimistic take which I will give (and then promptly duck as I expect this probably isn't the vibe of the conversation). Midjourney as a company was profitable enough and the founder retained enough control (he has no VC investors!) that the founder was able to take a big chance and build a first version of something that could actually be beneficial. Now we can all see together whether it was a good or bad idea...
The problem, in my opinion, is that doctors focus on curing patients. Any intervention that does not help with that is going to be ignored at best, actively opposed at worst. But gathering data, even if it leads to little in the way of useful clinical interventions on current patients, is valuable per se. You can use a million of those scans to train a foundation model for medicine. Maybe this will be helpful for *future* patients, even if it’s useless or even harmful for current patients.
The biggest issue with the bitter lesson fromy perspective is that it is premised on a misunderstanding of capitalism or at least of the dynamics of competitive markets. Even if you leave aside the question of whether more data or compute will improve your algorithm, the fundamental problem is that what matters is not whether more compute/data will solve your problem. The real problem is whether more compute/data will decrease the cost to be below having a human do the thing (taking into account the added cost of that added compute and data). Famously, both David Graber and Peter Theil both made this exact point about why there are "no flying cars like in the Jetsons" because workers wages are too low to incentivize new techmology development in the pages of the Baffler in like 2012.
The hope is always that data has positive network effects rather than diminishing marginal returns and the log relationship between training data and capability for LLMs sure seems like it should've settled this for even the most enthusiastic boosters of this line of thought, but noooo
Tech bros forget the people who die from natural causes do not die all of the sudden. The body shows symptoms of its own problems which should then receive medical attention.
Sounds like the first direct to consumer innovation where value of information (VOI) analysis can put claims of ‘more info is always better’ to the test. The conflation of data with information seems to be a sleight of hand in most of these conversations. I’m working on an information-theoretic, entropy based approach to answer the Q at a population level. We’ll see if MidJourney wants the value assessed beyond hype & hope. Great summary!
I would think a lot of the techno boosters are (mis)led by a belief that everything to do with compute/data exhibits emergent phenomenon with scale. Is this how people in Silicon Valley interpret The Bitter Lesson *jazz hands*?
But it's good to know that the radiologists are still around.
These are exactly the thoughts I have had around the hyped startup Neko Health, which performs a kind of full-body scan for preventive medicine. I guess we will see in a couple of years how that plays out. I am all for collecting data and information, but the maths involved does lead to a large number of unnecessary follow-ups, as you explain.
It's good to have a follow-up article on this. I was mid-jumping on the bandwagon when I realised what had occurred. This is still in what we would otherwise call "clinical development". Except it's exempt from therapeutic regulation under the FDA due to it being wellness-focused. Other countries have similar legislation.
Many jumped on the bandwagon because we thought this was exploitative- untargeted surveillance imaging which more than likely at current state, would cause problems. It's also not long after similar discussions on surveillance whole-body MRI (which has very few indications). But I think it deserves to be seen.
But, I also think it's a lesson for the health tech industry. Those may recall Google in 2017, "jumping the gun" and undermining their own excellent Streams project with Kings in the UK. Midjourney got excited and although they may not be subject to therapeutic regulation for this device development, they've experienced what will happen from early/overclaiming. Their business strategy may not depend on doctors, but they'd be far better to get them on side.
Do you know about liquid biopsies? I think they demonstrate that some of your negative claims are too strong, or maybe will be soon. About early cancer detection.
Everything we do in medicine has an upside and a downside. There are no free lunches. In the current discussion some possible downsides might be the societal cost of downstream testing for overwhelmingly incidental and benign findings. And the physical cost to the individual of complications from invasive downstream testing. Any finding from a screening test is most robust on the day it is done for the next day a tumor may arise. There may be a psychological cost if our lives become dominanted by repeated rounds of screening followed by diagnostic testing
Kind of feels like the same crowd that convinced itself that token-maxing was the right way to use AI is now trying to convince everyone that scan-maxing is healthcare.
"More information can make you sicker." You've articulated the real bitter lesson, for this century at least.
This debate revealed a lot about how information is handled by the medical system (LOL, the line about medical doctors and the medical establishment). For this particular technology, it may be right that it will not lead to better outcomes, but I contend that there is a lot of good information that is not obtained because of the way that information gets co-opted by the medical-industrial-legal complex, creating testing cascades, etc. To correct this, healthcare needs to move beyond Bayes rule toward decision theory. The other parties - industry, legal, etc - need to change as well.
I've got a more optimistic take which I will give (and then promptly duck as I expect this probably isn't the vibe of the conversation). Midjourney as a company was profitable enough and the founder retained enough control (he has no VC investors!) that the founder was able to take a big chance and build a first version of something that could actually be beneficial. Now we can all see together whether it was a good or bad idea...
The problem, in my opinion, is that doctors focus on curing patients. Any intervention that does not help with that is going to be ignored at best, actively opposed at worst. But gathering data, even if it leads to little in the way of useful clinical interventions on current patients, is valuable per se. You can use a million of those scans to train a foundation model for medicine. Maybe this will be helpful for *future* patients, even if it’s useless or even harmful for current patients.
The biggest issue with the bitter lesson fromy perspective is that it is premised on a misunderstanding of capitalism or at least of the dynamics of competitive markets. Even if you leave aside the question of whether more data or compute will improve your algorithm, the fundamental problem is that what matters is not whether more compute/data will solve your problem. The real problem is whether more compute/data will decrease the cost to be below having a human do the thing (taking into account the added cost of that added compute and data). Famously, both David Graber and Peter Theil both made this exact point about why there are "no flying cars like in the Jetsons" because workers wages are too low to incentivize new techmology development in the pages of the Baffler in like 2012.
The hope is always that data has positive network effects rather than diminishing marginal returns and the log relationship between training data and capability for LLMs sure seems like it should've settled this for even the most enthusiastic boosters of this line of thought, but noooo
Tech bros forget the people who die from natural causes do not die all of the sudden. The body shows symptoms of its own problems which should then receive medical attention.
Sounds like the first direct to consumer innovation where value of information (VOI) analysis can put claims of ‘more info is always better’ to the test. The conflation of data with information seems to be a sleight of hand in most of these conversations. I’m working on an information-theoretic, entropy based approach to answer the Q at a population level. We’ll see if MidJourney wants the value assessed beyond hype & hope. Great summary!
I would think a lot of the techno boosters are (mis)led by a belief that everything to do with compute/data exhibits emergent phenomenon with scale. Is this how people in Silicon Valley interpret The Bitter Lesson *jazz hands*?
But it's good to know that the radiologists are still around.
These are exactly the thoughts I have had around the hyped startup Neko Health, which performs a kind of full-body scan for preventive medicine. I guess we will see in a couple of years how that plays out. I am all for collecting data and information, but the maths involved does lead to a large number of unnecessary follow-ups, as you explain.
It's good to have a follow-up article on this. I was mid-jumping on the bandwagon when I realised what had occurred. This is still in what we would otherwise call "clinical development". Except it's exempt from therapeutic regulation under the FDA due to it being wellness-focused. Other countries have similar legislation.
Many jumped on the bandwagon because we thought this was exploitative- untargeted surveillance imaging which more than likely at current state, would cause problems. It's also not long after similar discussions on surveillance whole-body MRI (which has very few indications). But I think it deserves to be seen.
But, I also think it's a lesson for the health tech industry. Those may recall Google in 2017, "jumping the gun" and undermining their own excellent Streams project with Kings in the UK. Midjourney got excited and although they may not be subject to therapeutic regulation for this device development, they've experienced what will happen from early/overclaiming. Their business strategy may not depend on doctors, but they'd be far better to get them on side.
Do you know about liquid biopsies? I think they demonstrate that some of your negative claims are too strong, or maybe will be soon. About early cancer detection.
It kind of sounds like another Theranos to me😔
Everything we do in medicine has an upside and a downside. There are no free lunches. In the current discussion some possible downsides might be the societal cost of downstream testing for overwhelmingly incidental and benign findings. And the physical cost to the individual of complications from invasive downstream testing. Any finding from a screening test is most robust on the day it is done for the next day a tumor may arise. There may be a psychological cost if our lives become dominanted by repeated rounds of screening followed by diagnostic testing