He is an engineer from Guatemala, in his sixties. The Spanish accent has gone mostly quiet after decades in the States, but not entirely. Six months ago I did a radical prostatectomy on him. Localized disease, favorable risk, I spared the nerves. It went the way you hope these things go.

Today, sitting across from me in my clinic, his prostate-specific antigen (PSA) is 0.05. Which, if you are not in the field, is about as low as it gets. Some assays measure lower, but for practical purposes 0.05 at six months is what a good result looks like. It is a number you would, in any other setting, celebrate. He is not celebrating. And the fact that he is not celebrating is, more or less, the whole point of this essay.

Let me back up.

When he first walked into my clinic, before any of the surgery, he brought a folder. An actual folder. With tabs. The kind of folder that suggested, accurately, that this patient had been preparing for this visit for considerably longer than I had. He spread it across my desk and walked me through its contents as if I were the one being briefed, not him. His PSA trend from the last three years. The dimensions, in millimeters, of the lesion on his multiparametric MRI. His Gleason score. His biopsy cores, positive and negative, in a small table he had made himself in what looked suspiciously like Excel. I have, for the record, never before been briefed on my own patient’s pathology by the patient. It is a humbling experience. I recommend it.

He looked up from the folder, very politely, and said, “Doctor, I am an engineer, I like my numbers. Can you walk me through what you think is happening, in numbers, before we talk about anything else?”

So I did. I pulled a blank sheet of exam paper and I drew him a picture. Bladder, urethra, a small triangle for the prostate. I wrote numbers next to each of them. Cancer numbers. Continence numbers. Erection numbers. I remember the numbers I gave him that afternoon. He remembers them too. That, really, is the whole problem.

Today, six months out, he has his 0.05, and he also has questions. What is the lower limit of detection on this particular assay? How confident am I, in a percentage, that 0.05 means what we think it means? What does it mean, statistically, that his final pathology noted a one-millimeter focus of extraprostatic extension, and how does that interact with the 0.05 on the screen today? If the number rises next time, even slightly, at what threshold do we act?

These are good questions. These are not the questions I am used to getting in clinic. The questions I am used to are closer to, “Doctor, is my cancer back,” which, for all its weight, is at least a single question with a shape I have learned to answer. His questions are harder. His questions are the kind where the honest answer begins with “we don’t entirely know,” and then goes on for the next twenty minutes, during which I am acutely aware that my continuing medical education (CME) credits on prostate cancer biology had better be up to date, because if they are not, this man will notice.

This is the part of the job that nobody trained me for.

I am a urologic oncologist. Early faculty, which is a polite way of saying that I am old enough to technically run a clinic and young enough that my mother still occasionally asks, via a phone call I always take, whether I have considered going into something less stressful. My clinic, like most clinics in this field, is mostly men in their sixties and seventies, with folders and wives and questions. Mine today brought all three.

What I was trained for, to be blunt, is to cut on these men. Cutting is a craft, and a craft can be taught. You drill. You simulate. You watch someone do it a hundred times and then, one day, you do it yourself, and a few hundred repetitions later you do it well. There is a book for all of it. There is a fellowship. There are videos. There are, God help us, podcasts.

The clinic, as it turns out, cannot be simulated.

I have another patient, a retired school principal in his early seventies. His prostatectomy was done two years ago, by someone else, at another hospital, back when I was still in fellowship. He came to me a few months back, after his original surgeon moved on. His cancer is fine. His PSA is undetectable. That is not what he comes in to talk about. He comes to talk about the Depends. He has been wearing them under his slacks, every day, for two years. He still leaks when he laughs too hard. He stopped dancing with his wife at weddings about a year ago, because he does not trust his own body for the three minutes of a slow song. His cancer, objectively, is cured. His life, he tells me, is not quite his life.

There is no trial that answers the question he is really asking me, which is whether it might have been better to have done nothing, and whether the cancer, left alone, would have killed him more kindly than the surgery has. There is no algorithm. There is only him, and me, and the exam room, and the fact that he is asking.

The thing I have been trying to find words for, and I am still not sure I have them, is that the branches of medicine where you actually face the patient are a different animal from the rest of medicine, in a way I don’t think is obvious from outside. I have friends in radiology. Pathology. Anesthesia. Critical care. I love them. Some of them are, by any honest accounting, smarter and braver and more technically gifted than I am. The anesthetist who pulls a difficult airway out of the fire at 3 a.m. is doing something I could not do in a lifetime, and also has, as far as I can tell, personally renegotiated his relationship with the concept of sleep. The ICU physician running a code on a dying septic patient is carrying a weight I have not earned the right to speak about.

But the weight is shaped differently.

In the dark reading room, the slide tray, the OR, the ICU bay, the patient is covered, draped, sedated, or reduced to a line on a monitor. The person is there. The face is not. And the face, I have come to think, is a different thing from the person. The face is the part that looks back.

In clinic, the face is always in the room. It is in the room today, and it will be in the room again in three months when we check the number, and again three months after that. A prostatectomy, it turns out, is not a procedure. It is the opening sentence of a long conversation between two people who did not exactly choose each other.

Here is what else nobody warned me about, and I wish someone had. The complication is not always yours.

A different patient of mine, one I had referred for adjuvant radiation, a patient for whom that referral was objectively the right call given his pathology, developed a rectourethral fistula. A small, terrible communication between two organs that were never meant to meet. Not my beam. Not my plan. Not my hand. He came back to me anyway.

He sat in the same chair, across the same desk, and he said, without any accusation, which I think was actually the hardest part, “Doctor, what happened?”

You cannot, in that moment, explain to him the division of labor in modern oncology. You cannot say that was a different team. You cannot hide behind the referral letter. He came to you. His life was entrusted to a constellation of people, and from where he was standing, you were the brightest star in that constellation, and something has gone wrong, and you are the face of it now. This is also the job. It is the particular weight of being the person a patient looks at when he wants the truth about what his body has become. You may not have caused the thing. You are, nevertheless, in charge of the story.

I don’t think there is a real equivalent of this in the branches where the patient is not in front of you. I am sure there are burdens in those branches I don’t know about. But those burdens are carried, mostly, in rooms full of other professionals, in a language those professionals share. They are not carried across a desk from a man who is trying, and failing, not to cry.

People sometimes ask me, usually on a padel court, usually between games, whether I am worried about AI taking my job. I say no, and I say it too quickly, and then I feel obligated to explain, because saying no too quickly tends to be received, even on a padel court, with the particular look people give when they suspect you of being in denial about something.

Here is the honest version.

I think, within my career, a machine will read an MRI better than I can. I think it will predict recurrence better than I can. It will probably, eventually, read a Gleason slide better than most pathologists. I welcome all of this. I would rather be a slightly worse diagnostician than my patient’s best chance, and if the algorithm is better at the parts of the work where being right matters more than being human, I want the algorithm on those parts. Let me have the parts where being human matters more.

Because the chair across from me is not a computational problem. It is a moral one. The engineer from Guatemala, despite every appearance, is not actually asking me for a prediction. He is, after all the numbers and all the assay thresholds and all the decimal places, asking me what this means for him. For his wife. For the way he gets to live the next ten years. Even if a model could generate the prediction more accurately than I can, it cannot generate the answer, because the answer is not a number. The answer is a conversation between two people, one of whom is frightened and one of whom is responsible, and the medium of that conversation is trust.

You cannot automate trust. You can only earn it, slowly, visit after visit, by the boring analog work of being the same person in the same chair every single time.

I want to be careful here not to make this sound noble. Most of what happens in clinic is a long accumulation of imperfect human gestures. You forget a wife’s name until the patient mentions her, and then you are visibly, awkwardly grateful. You say the wrong word on a day you have not slept. You usually do not find out which of those small moments mattered until years later, if ever.

But the character of the work is defined by the fact that the gestures are yours and the face they land on is his. There is no drape. There is no screen. He sees you. You see him. Something happens in the space between the two chairs that does not make it into any note I later dictate, and that thing, I think, is the actual point of the specialty.

A surgeon who does not understand this can still operate. He can still cut beautifully. He’ll still get the margin. But over a career, the mismatch between what he was trained for and what the work actually asks of him will wear him down, and he will become either a cynic or a recluse. (Those are the two options, in my observation so far. The cynics are easier to spot. The recluses tend to have nicer operating rooms.) Either way, he retreats into the OR, where the patient is asleep and the team speaks his language and the outcomes can be measured in milliliters and minutes.

I recognize the pull of that retreat more days than I would like to admit. The clinic is where I find out, every week, whether I am going to stay out of it.

The engineer does not cry. Engineers, in my limited and unscientific experience, do not often cry in my clinic, though I have been wrong about this enough times that I have stopped assuming. He listens. He writes things down. He asks, at the end of the visit, whether he can email me during the week if more questions come up, and I say yes, because he is going to email me whether I say yes or not, and we both know this. I tell him we will repeat the PSA in three months. I tell him that if the number rises, we will have that conversation then, and not before. He tells me, politely, that he is going to worry about it anyway. I tell him that is reasonable, because it is.

He stands. He puts on his jacket. He does the small, tidy, dignified thing men of his generation do when they are trying not to let you see that they are holding something. He shakes my hand. He thanks me, which is almost unbearable. He walks out.

The room is quiet. The 0.05 is still on the screen. A single decimal, two places in. A small verdict, in his favor.

I sit for maybe thirty seconds before I call the next patient. In those thirty seconds I do the thing every clinician I admire has, in one form or another, told me you have to do if you are going to do this for a long time. I let the weight sit. I don’t push it away. I don’t pretend it is lighter than it is. Even a good number has a weight, when it is attached to a person. I acknowledge that someone has just passed through my hands and left a mark on me, as I have, I hope, left something useful on him. Then I stand up and call the next name.

If you are in one of the branches where the face is always in the room, urology, oncology, obstetrics, palliative, any specialty where people come back and back and back, I want to say something to you. The weight you are carrying is not evidence that you are bad at this. It is evidence that you understand what this is. The day the face stops mattering to you, the day you can hand somebody their 0.05 and feel absolutely nothing, is the day you probably should not be the one handing out 0.05s.

If the OR is the longest breath, the clinic is what you do with your lungs afterwards. Above the surface. Face to face. Eye to eye.

Stay in the chair.

Archan Khandekar is a urologic oncologist.