IBM’s Watson Health is sold off in parts

There already was a tiered system, with rich people being able to buy concierge medicine and getting preferred treatment based on who knows who on the hospital's board or if their name is on a wing of the hospital.

The change now is a more visible and more granular price segmentation.

1. A basic solution shipped and made a ton of money e.g. Ads, Search, recommendations etc.

2. It is financially feasible to have a dedicated team(s) make small incremental progress on these solutions. Even very small gains are beneficial.

3. The business perceives a threat if they fall behind in this area.

The thing is that the gains on the basic solution (heuristics, off the shelf pre-trained CV model, open voice recognition) are pretty small, and if the threat of others making progress goes away - the inferred value of further investment will probably vanish as well.

Other applications which put the AI in the driver's seat (sometimes literally) seem far from production - or if they do work, then they work reasonably well using an alternate approach from what you might expect.

Hence the purpose of managed care organizations (MCOs, health insurance companies) employing people to approve and deny (or design systems that approve or deny) payment for unnecessary tests, consults, and scans. And in a taxpayer funded system, the government employs people to performs the same roles.

Perhaps this indicates a major gap in your visibility into and understanding of modern medical practice?

> My mom worked for a GP for about 20 years, and it seemed to me that most of what made that guy a doctor was bedside manner + being able to remember a lot of things.

"being able to remember a lot of things" -- Yes, absolutely. This is an area where AI-assisted decision support (like Watson) could be, and I hope will be, extremely valuable.

OTOH, you also rightly recognize bedside manner -- as anyone who has been a patient or a patient's family caregiver will attest, this is an essential component of core, and I think likely also the healing process. We won't get this from a machine until there is a true general purpose AI, at which point I expect an AI singularity anyway.

But consider that there are other factors of the practice of general medicine that you haven't even touched upon. For example, liability: In the office you envision, where Watson/AI makes medical decisions and these are executed by other roles like nursing, where does ultimate responsibility (and legal liability) lie? Remember that Watson outputs probabilities -- suppose the following:

Chance of disease X: 89% Chance of disease Y: 10% Chance of disease Z: 1%

If disease X, intervention A has an 80% probability of success. However, if disease Y, intervention A has an 80% probability of harm.

(before you object that this is contrived, this represents a realistic situation I encountered recently, where intervention A is high dose steroids)

Now, will we put the onus on the patient to select an intervention? I don't think that would be very popular. While I certainly do not advocate paternalism, when faced with difficult decisions quite many people openly defer to their physician.

> But GPs often make astounding amounts of money

I suppose this could be true for some definitions of "astounding," but it's generally accepted that GP/"primary care physician" pay is essentially the lowest of all specialties , which is a major contributor to lack of access to primary care in industrialized countries (while you'll not have a hard time finding a private pay dermatologist, for instance)

ANd on top of this, we are posting on HN where a mid level software engineer total comp can EASILY outpace the average US primary care physician salary.

> while leaning heavily on their staff to actually handle patients and keep the business running.

Do not all professionals and business executives rely on highly trained staff as force multipliers? This is a fundamental principle of the advancement of human economies. It is grossly inefficient to operate with individuals as "jack of all trades" when they can instead each become specialized to support a bigger or broader goal.

By "leaning heavily on their staff to actually handle patients and keep the business running" are you suggesting that the primary care physician should be performing check-in, insurance verification, rooming, vitals measurement, blood draw, medication administration etc? That is a certain recipe for massively decreased throughput and shortages/decreased access to primary care.

> I thought it could help drugs get a little cheaper too, because there wouldn't be any point in the pharma companies sending out salespeople to do lunch seminars to convince the GPs to prescribe this or that drug (this still happens).

This has been massively curtailed for 20+ years, at least in the US. I am not sure about other countries. But overall I think this is a seriously minor portion of the (exorbitant) price of medications in industrialized countries.

In any case, I doubt it would make much of an impact on utilization in most contexts, as insurance/health plans/prescription benefits have already implemented fairly strict guidelines-based formularies and coverage tiers (again, at least in the US -- I can't speak to other industrialized countries, although I expect they are similar or even more strict)

(edit: another poster points out all the direct to consumder drug advertising -- I agree - this probably has a much bigger influence in 2022; ad budgets are absolutely insance)

> Maybe this will still happen, but it doesn't seem imminent anymore.

Here we agree. I am certain we'll see AI-assisted physician decision support, but (a) the physician won't go away and (b) I think it'll be an unfortunately long ways in the future.