What Robotaxis Mean for Healthcare

You may have seen that last week the California Public Utilities Commission (CPUC) gave approval for two companies to operate self-driving taxicabs (“robotaxis”) in San Francisco, available 24/7 and able to charge fares.  Think Uber or Lyft but without drivers.

Meet the future. Credit: Waymo LLC

It has seemed inevitable for several years now, yet we’re not really ready.  It reminds me, of course, of how the future is coming fast for healthcare too, especially around artificial intelligence, and we’re not really ready for that either.

The two companies, Cruise (owned by GM) and Waymo (owned by Alphabet) have been testing the service for some time, under certain restrictions, and this approval loosens (but does not completely remove) the restrictions. The approval was not without controversy; indeed, the San Francisco police and fire departments,  among others, opposed it. "They are failing to regulate a dangerous, nascent industry," said Justin Kloczko, a tech and privacy advocate for consumer protection non-profit Consumer Watchdog.  

The companies brag about their record of no fatalities, but the San Francisco Municipal Transportation Agency has collected almost 600 “incidents” involving autonomous vehicles, even with what they believe is very incomplete reporting.  “While we do not yet have the data to judge AVs against the standard human drivers are setting,” CPUC Commissioner John Reynolds admitted, “I do believe in the potential of this technology to increase safety on the roadway.”

Credit: Cruise LLC

I’m willing to stipulate that autonomous vehicle technology is not quite there yet, especially when mostly surrounded by human-driven vehicles, but I also have great confidence that we’ll get there quickly, and that it will radically change not just our driving but also our desire for owning vehicles.

One of the most thoughtful discussions I’ve on the topic is from David Zipper in The Atlantic. He posits:

A century ago, the U.S. began rearranging its cities to accommodate the most futuristic vehicles of the era, privately owned automobiles—making decisions that have undermined urban life ever since. Robotaxis could prove equally transformative, which makes proceeding with caution all the more necessary.

A century ago, cities didn’t really have speed limits, vehicle-only streets, parking spaces/lots/garages, and certainly not freeways cutting through city neighborhoods.  He notes: “More than half of the land in many downtowns is used to move and store motor vehicles, occupying space that could otherwise accommodate housing, retail, playgrounds, and parks.” 

Mr. Zimmer warns that wider use of autonomous vehicles could lead to more miles driven, and much less public transit used.  He cites a 2018 quote from Frank Chen, of Andreessen Horowitz:   “We don’t understand the economics of self-driving cars because we haven’t experienced them yet. Let’s see how it plays out.”  But waiting to see how the future plays out often means not really planning for it.

If we were certain that in, say, 10-15 years, most vehicles would be autonomous, or at least could communicate/coordinate with each other, we’d be making different investments in things like smart roads, speed enforcement cameras, traffic police, or – you guessed it -- parking.  Some of those investments take years to bring about, so we’d better lay out bets down soon if we want to be ready. We can’t even seem to get our heads around a future with largely electric vehicles (EVs) – e.g., where are all the charging stations? – so preparing for autonomous vehicles seems like a future we’re not thinking enough about.

Which leads me to healthcare. 

For example, one of my pet peeves about the discussions of using AI in healthcare are the ones about how they’ll help with “paperwork.”  AI will take care of physician notes!  AI will handle all the prior authorization requests!  AI will help fill out all those patient forms!  All of those, and many more, certainly reflect 2023 pain points in healthcare, but they’re missing the future.  AI should not do the paperwork, hiding it from us; it should help us eliminate the need for it.  Don’t use AI to make the system work faster, use AI to make the system more efficient and effective.

Don't do this. Credit: Bing

In an AI world, for example, there should never be a question about a patient’s eligibility or available benefits; we’d know those in real time. Similarly, I don’t think there will be a role for claims processing as we know it; instead, we’d have real time adjudication and payment.  Don’t even get me started on what we now ludicrously call “price transparency;” AI will take all the guesswork out.

Buildings are healthcare’s version of cities’ parking lots.  We require a lot of them, be they hospitals, medical office buildings, or others, because patients usually have to be in such places to get care. They’re also like parking because much of what patients do in them is wait.

The future healthcare system is going to be much more distributed. Care will shift from being given in a place where health care professionals are to where you and your devices are, with those devices omnipresent, connected, and smart.  They will sometimes but not always interact with human clinicians.  So which buildings go, when? 

Healthcare will also be much more proactive and predictive. Impressed with what your Apple Watch can monitor?   That’s nothing compared to the monitoring we’ll have available in the near future, including those long-predicted nanobots swimming around in us. We’ll know how we’re doing, we’ll know how we’re trending, and in many cases we’ll know well in advance when it is time to take action.

AI will also help us make better predictions about which people are going to need expensive care, when, which will have dramatic impacts on how health insurance works and when it is actually needed.  Plan for that future.

My last warning, which I’ve touched on before, is how AI impacts payment.  Right now some argue we should treat AI as medical devices, regulating and paying along the same lines. Others want more CPT codes for AI, building on our existing mess of that system.  Neither of those are anticipating the future; they’re trying to lock in the present.

You're pricing AI wrong. Credit: Bing

We should be thinking about pricing AI as the technology that it is, which should get cheaper yet more powerful as time goes by. If the AMA, through its CPT Editorial Panel, gets to decide how AI is priced, God help us and our healthcare system. 

We’re not ready for self-driving cars, or for AI in healthcare, which is understandable given that both have a long way to go.  But the future comes at us fast, so if we’re not preparing now we’ll be too late.