.svg)
A week ago, I wrote about getting trapped in a Waymo on the way to the CAP annual meeting. That ride taught me something important about trust, control, and what it feels like to depend on a technology when you are no longer the person demonstrating it.
So naturally, when I got to San Diego for Pathology Visions 2026, I decided to do the obvious thing. I got in another Waymo.
This time, my colleague Jeff was along for the ride. And while the experience was a lot less dramatic, it may have been even more useful.
Because this ride raised a different set of questions. Not about whether the technology works. About whether it is ready to win.
Safety first
At one point during the ride, we were traveling on a four-lane road. A pedestrian was standing near the edge of the street and appeared to be considering crossing. There was no crosswalk. No intersection. He had not actually stepped in front of us.
But the Waymo saw him. And it stopped. It waited. The pedestrian crossed, and then we continued.
There are plenty of jokes you can make about autonomous vehicles being overly cautious, but I actually found that moment reassuring. The system clearly had a bias. Safety first.
That principle translates directly to digital pathology and AI. Healthcare is not a place where you earn extra points for being aggressive with uncertainty.
The technology needs to know when to proceed. It also needs to know when not to. As we add AI into pathology workflows, that matters enormously.
The best AI will not simply be the system that produces the fastest answer. It will be the system that knows its limits. That recognizes uncertainty. That asks for human judgment when appropriate. That protects the patient first.
I think most of us would agree with that. But safety alone is not enough.
The route still matters
Our Waymo ride from the hotel to the conference should have been a relatively short trip. Instead, it became a roughly 25-minute adventure through side streets and indirect routing.
Nothing went wrong. We got there. But it was hard to ignore the fact that the route made very little sense to two humans sitting in the car.
And this was not the first time I had seen it. The same thing happened on our previous Waymo experience. The system found a route. It just was not the route we would have chosen.
That creates another important parallel with digital pathology. A new technology can be safe. It can work. It can even be impressive. But if it forces the user through a workflow that is slower, more complicated, or less intuitive than the one it replaces, eventually the novelty wears off.
And novelty definitely matters. People noticed our Waymo. Several people took pictures. We rolled down the window and waved. It was fun.
It is undeniably cool to sit in a car with nobody driving it. But being cool is not enough. Especially when you are running a pathology department.
A department chair does not get rewarded because the technology looks futuristic. A pathologist does not get extra time in the day because the software is impressive. A lab does not improve because people want to take pictures of the interface.
Eventually, the technology has to make the work better. It has to improve capacity. It has to reduce friction.
It has to help people reach the destination with less effort, not simply take them on a more interesting route.
That is the standard.
Then there is the cost
The final thing that caught my attention was the price. Our Waymo ride cost roughly twice what a comparable Uber ride had cost earlier in the week.
Again, that is understandable. New technology often carries a premium. Early adopters are willing to pay more for access to something new, interesting, and different.
That can work for a while. But it is not how technology reaches scale. Eventually, the economics have to work.
That is true for autonomous vehicles. And it is absolutely true for digital pathology and AI. There has to be a return on investment.
The technology has to make something better. Or faster. Or less expensive. Ideally, it does more than one of those.
If the new technology costs more, takes longer, and produces essentially the same outcome, enthusiasm eventually runs into economics.
That is not cynicism. That is how organizations make decisions. Hospitals have budgets. Pathology departments have budgets. People are accountable for capital, staffing, throughput, turnaround time, quality, reimbursement, and increasingly tight margins.
The argument for digital pathology cannot simply be: “This is the future.”
And the argument for AI cannot simply be: “Look what the algorithm can do.”
Those statements may be true. They are just not enough.
The standard has to get higher
As I reflect on our time at Pathology Visions this week, that is probably my biggest takeaway. Digital pathology is clearly moving forward. AI is clearly going to be part of the future of pathology. The technology is becoming more capable. The excitement is real. But now we have to hold ourselves to a higher standard.
Safety first. Absolutely.
But also better workflows.
More capacity.
Faster answers where speed actually matters.
Lower friction.
Better use of scarce pathologist time.
Improved access to expertise.
More consistency.
And ultimately, an economic case that makes sense.
The technology has to earn its place.
That may take time. Most important technologies do. And I have spent enough time in healthcare to know that adoption almost never happens as quickly as the people building the technology expect.
But I also believe this strongly: With the right technology, the right implementation, and the right workflow design, pathology will get there.
We will create systems that are safer. Faster. More capable. And economically sustainable.
The challenge for everyone building in this space is to make sure we do not confuse “new” with “better.”
Our second Waymo ride got us where we needed to go. It was safe. It was interesting. It was definitely cool. But it took longer. And it cost more.
That is probably fine for an emerging technology. It is not the standard for one that wants to become indispensable. Digital pathology and AI should hold themselves to the same test.
The future cannot just be different. It has to be better.