Offcall Physician Spotlight: Meet Dr. Stav Devons-Sberro

Have a response for Dr. Devons-Sberro? Reply to this post directly — she’ll personally read every message. Also, let us know who we should feature next by replying directly to this post!

1. Stav, what’s one task you’ve fully handed off to AI that you’d never take back? Having it review the grammar and wording of my formal writing. As someone who is not US-based, and for whom English is only my father tongue :), there was always a small but very real anxiety around professional correspondence. I always worried I might miss something — a phrase, a convention, a commonly used term — and somehow come across as less professional than I actually am. In the past, when I had to correspond formally with professional or official bodies, I would check everything 10 times. Today, it has honestly become a non-issue for me. In that sense, AI has been a very meaningful barrier-remover. Now you’ll have to guess how much of this answer was written by me and how much by AI. (Hint: I have a much better sense of humor…)

2. Are you worried about deskilling — that the next generation won’t develop the intuition you did because the model handled it? I’ll answer like a good doctor: it depends :) Yes, there will be tasks that someone starting medical school today may never really know how to do without AI. In the same way that no one today listens directly to a patient’s chest without a stethoscope — and even that, if we’re being honest, has often become an echo. And while AI may not replace physicians, it will absolutely replace some of the tasks we currently do. And that is not necessarily a bad thing. We should be honest: There are things artificial intelligence simply does better than human intelligence, especially when it comes to processing very large amounts of information. For example, reviewing 10 years of trends in a patient’s blood tests. I think much of the future of medicine, like many other professions, will be about human-machine collaboration.

But the first thing this should change is how we teach students. Skills like having a phenomenal memory used to be part of what made some doctors exceptional. Today, that matters a little less. But skills like critical thinking, clinical judgment, and the ability to doubt are becoming even more essential — especially when ChatGPT sounds more confident than a medical student on their first day on the ward.

So the way we teach and assess medical students has to change. For example, in a clinical vignette, instead of only asking, “What is the diagnosis?”, we could give the case together with an AI-generated diagnosis and ask: “Which parts of the AI’s answer do you trust?”, “What information is missing?”, “What red flags may have been missed?”, and “Where is there a risk of automation bias?”

3. Will AI solve the physician shortage or quietly make it worse? I don’t think AI’s impact on the physician shortage will be uniform across all areas of medicine. That’s part of the problem with talking about “medicine” as if it were one single thing.

In preventive medicine, AI opens up opportunities at a scale that is hard to overstate — from predictive models for almost every disease imaginable, to the ability to detect subtle findings in imaging or pathology, identify trends in lab results, or generate personalized recommendations based on guidelines and a patient’s own medical record. We may be able to practice preventive medicine in a way that is truly remarkable, and maybe even reduce the burden of some diseases. Spoiler: this is somewhat close to what I do in my day job as a physician at Clalit Innovation.

On the other hand, the very “ease” of accessing some AI-based healthcare services may actually increase demand and make the shortage worse. And yes, some physicians may lose a sense of challenge or meaning and leave the profession. For many others, AI may significantly change the day-to-day work of being a doctor. But in that sense, physicians are certainly not alone. AI is changing the entire job market. Not exactly comforting, but at least we’re in good company.

Download to join the waitlist

Medicine is complex enough.
Referrals shouldn't be.

Send and receive referrals, build wealth, and grow your physician community with Offcall.

Download on the

App Store

GET IT ON

Google Play

4. What’s something you used to judge other physicians for that you now completely understand? I used to be more judgmental of physicians who set very firm boundaries around patients’ access to them. In medical school, we were taught to admire doctors who gave everything to the profession: the ones who shared their personal phone number, were always available, and sacrificed huge parts of their own well-being for their patients. We saw that as the highest form of dedication.

So when we met a physician who wouldn’t stay another 10 minutes in clinic for a patient who arrived late, or who insisted that patients contact them only through the office, it felt strange. Maybe even cold. Now I completely understand it. This work is deeply draining. And we should also be honest, many patients do not always appreciate that kind of availability and sacrifice. Some even take advantage of it. Those boundaries are not necessarily about arrogance or not caring. Sometimes they are simply the very reasonable expression of an adult who understands their priorities in life.

5. What did you spend money on early in your career that absolutely wasn’t worth it? Books. Especially the kind people tell you that you “absolutely must have.” In my medical school, everyone bought Pocket Medicine almost obsessively before the internal medicine rotation. I think I opened it maybe twice. I just didn’t connect with the format, and it didn’t feel like a good learning source for me.

Of course, today everyone has a tablet, but this is not just about paper versus digital. One of the best things I learned in medical school was how I personally learn best — what actually helps me process and understand information. And now, with the level of personalization we can get from technology, that becomes even more powerful. Learning how to learn has always been an important skill. Right now, maybe more than ever.

6. What’s something physicians are quietly worried about that the public doesn’t know about yet? How long can we hold on? “How long can we keep up like this?” I think the public knows that physicians are burned out, and that the profession has become increasingly difficult. But I’m still not sure people understand just how much.

I know quite a few physicians who are genuinely afraid of being “the last ones to leave.” Every few days, you hear about someone else who left, switched to a different role, reduced their clinical hours, or found a way to be a little less inside the system. And all the talk about technology eventually replacing us does not exactly help here. There is this quiet but very real fear among doctors of missing the train. Of being the last one still standing on the ward after everyone else has already understood something you haven’t, that maybe they should have left earlier.

7. If you had a magic wand for one thing in medicine — not your whole career, just one thing — what would you fix tomorrow? I would fix the residency Match — or more precisely, the way we decide who gets into which specialty, and where.

Not only because residency itself is hard — though it absolutely is, especially when it often overlaps with the years of building a family, and unsurprisingly, that burden does not fall equally on women. But because residency selection exposes a deeper problem: we still don’t really know how to define what makes someone a good doctor. We know how to measure exam scores, which often test memory, speed, and recall. We know how to conduct interviews in an artificial, high-pressure setting. We know how to be impressed by recommendation letters, CVs, and someone’s ability to present themselves well. But most of us know that this is not necessarily what makes a good physician.

So we do not have good enough tools to evaluate candidates for residency. And when the tools are not good enough, opportunities do not always go to the people best suited for them. The amazing part is that we copied this exact problem almost perfectly into AI.

We started by testing models like medical students: multiple-choice questions, then something a bit closer to OSCEs, then clinical simulations. And then we get frustrated, rightly, when tech companies claim that because a model answered more questions correctly than physicians, it can replace them. We say “wait, that tells us almost nothing about how it would function as a doctor in the real world!”

And that is completely true. But maybe it is also a mirror. If we still don’t really know how to measure what makes a good human doctor, maybe we shouldn’t be so shocked that we are also measuring AI doctors in all the wrong ways.

2025 PHYSICIANS AI REPORT

Access the Complete Report

✓Complete quantitative breakdown of what physicians really think about AI

✓Strategic implications for healthcare organizations and AI companies

✓Sentiment analysis of physician attitudes about AI and the future

Download full report