By KIM BELLARD

Last week I was on a fun podcast with a bunch of people who were, as usual, smarter than me, and, in particular, more knowledgeable about one of my favorite topics – artificial intelligence (A.I.), particularly for healthcare.  With the WHO releasing its “first global report” on A.I. — Ethics & Governance of Artificial Intelligence for Health – and with no shortage of other experts weighing in recently, it seemed like a good time to revisit the topic. 

My prediction: it’s not going to work out quite like we expect, and it probably shouldn’t. 

“Like all new technology, artificial intelligence holds enormous potential for improving the health of millions of people around the world, but like all technology it can also be misused and cause harm,” Dr Tedros Adhanom Ghebreyesus, WHO Director-General, said in a statement.  He’s right on both counts.

WHO’s proposed six principles are:

• Protecting human autonomy
• Promoting human well-being and safety and the public interest
• Ensuring transparency, explainability and intelligibility 
• Fostering responsibility and accountability
• Ensuring inclusiveness and equity 
• Promoting AI that is responsive and sustainable

All valid points, but, as we’re already learning, easier to propose than to ensure.  Just ask Timnit Gebru.  When it comes to using new technologies, we’re not so good about thinking through their implications, much less ensuring that everyone benefits.  We’re more of a “let the genie out of the bottle and see what happens” kind of species, and I hope our future AI overlords don’t laugh too much about that. 

As Stacey Higginbotham asks in IEEE Spectrum, “how do we know if a new technology is serving a greater good or policy goal, or merely boosting a company’s profit margins?…we have no idea how to make it work for society’s goals, rather than a company’s, or an individual’s.”   She further notes that “we haven’t even established what those benefits should be.”

Ms. Higginbotham isn’t specifically talking about healthcare, but she could be.  We can’t really agree on what a healthcare system should and shouldn’t do, much less one augmented by A.I.  It’s no wonder that our first generations of A.I. in healthcare are confused.

The example that I’ve been using for years is that we can’t even agree on how human physicians seeing patients in other states via telehealth should be licensed/regulated, so how are we going to decide how a cloud-based healthcare A.I. should be?   

Carissa Véliz has an idea.  Writing in Harvard Business Review, she suggests that the FDA test AI like it does prescription drugs or medical devices, using randomized control trials to prove validity and efficacy.  I’d feel better about that if we didn’t already have a lot of history of that process taking too long, being swayed by non-data driven factors (e.g., Aduhelm), or being frequently circumvented. 

It gets worse.  Christopher Mims just wrote about how AI is moving from the cloud to edge devices (like your phone or home appliance).  Edge computing is going to be a big part of our future, including healthcare, but, as computer science professor Elisa Bertino pointed out to him, how can anyone certify/regulate AI that is evolving on its own, in the real world?  It won’t necessarily resemble the A.I. that it started out as; it’s going to depend on the data/inputs it receives. 

Mr. Mims also warns: “Modern AI, which is primarily used to recognize patterns, can have difficulty coping with inputs outside of the data it was trained on.”  Oh, boy — it’s going to run into a lot of that with health care.  People are messy, so to speak, and a lot of that mess impacts their health.  A.I. better be ready to deal with it. 

AI is going to evolve much more rapidly than other healthcare technologies, and our existing regulatory practices may not be sufficient, especially in a global market (as we’ve seen with CRISPR).  Not to be facetious, but we may need AI regulators to oversee AI clinicians/clinical support, just as we may need AI lawyers to handle the inevitable AI-related malpractice suits.  Only another black box may be able to understand what a black box is doing.

I worry that we’re thinking about how we can use A.I. to make our healthcare system do more of the same, just better.  I think that’s the wrong approach.  We should be going to ground principles.  What do we want from our healthcare system?  And, then, how can A.I. help get us there? 

For example, we should want that everyone has access to affordable health care – when they need it, where they prefer it.  That health care should tailored to the individual, including genetics, environment, and socio-economic status, and should be based on solid evidence.  That all sounds like a list of the usual platitudes, but none of it is currently true.  How can A.I. help make it true, or, at least, truer?

If A.I. for healthcare is a better Siri or a new decision support tool in an EHR, we’ve failed.  If we’re setting the bar for A.I. to only support clinicians, or even to replicate physicians’ current functions, we’ve failed.  We should be expecting much more.

E.g., how can we use A.I. to democratize health care, to get advice and even treatment in people’s hands?  How can we use it to help health care be much more affordable?  How can A.I. help diagnose issues sooner and deliver recommendations faster and more accurately? 

In short, how can A.I. help us reorient our health care from the healthcare system that delivers it, and the people who work in it, to our health?  If that means making some of those irrelevant, or at least greatly redefining their roles, so be it. 

Right now, much A.I. work in healthcare seems to be focused primarily on granular problems, such as diagnosing specific diseases.  That’s understandable, as data is most comparable/available around granular tools (e.g., imaging) or conditions (e.g., breast cancer).  But our health is usually not confined within service lines.  We need more macro A.I. approaches.  We might need A.I. to tell us how A.I. can not just improve our healthcare but also to “fix” our healthcare system.  And I’m OK with that. 

Kim is a former emarketing exec at a major Blues plan, editor of the late & lamented Tincture.io, and now regular THCB contributor.