AI and money: How tech could upend the doctor income relativity divide

I know doctor-doctor couples with big income disparity. Let’s imagine one in Calgary: She’s a pediatrician (average gross billings $443,426) and he’s an ENT (average gross billings $715,287). I’ve never asked couples I know like this how they think about income relativity and equality but I wonder if they ever have fights that mirror the frustration this drives in the profession?“How can you not see that the pay differential is unfair and is driving burnout and the family doctor shortage?” she said, taking her plate to the dishwasher. “Doctors who ‘do things to patients’ have had tech improve—and that lets you do procedures faster but that isn’t available to those of us who ‘talk to and manage patients.’”He frowned. “You don’t understand the nature of my work. First, looking at gross billings isn’t fair. We have massive clinic overhead costs! Plus, for many surgeons, the work takes an intense physical toll and there are very real liability issues in the OR. And, well, supply and demand works here: Many of these surgeons have scarce technical skills.”“It’s a kind of ‘pink tax,’” she said. “As a specialty becomes more female-dominated, its median pay growth slows. Patients often expect women docs to spend more time with them and provide more emotional support which is challenging in a fee-for-service environment.”And so it goes. . . . These issues have driven immense tension within provincial medical associations for a long time. I remember covering fee and income relativity challenges years ago in Ontario just after PACS and phacoemulsification had dramatically changed how much time some doctors spent doing imaging and cataracts.But now another round of tech changes is hitting the profession. AI by specialtyThe conventional wisdom on AI’s impact is this: It is the cognitive specialties, where the work involves lots of synthesizing histories, analyzing data and documenting plans, that are going to see the most efficiencies and changes in daily workflows from AI. The surgical and procedural specialties, dealing with the physical reality of human bodies, the argument goes, are to see less disruption. Indeed, I was having a conversation with my main AI, Gemini, as I thought through all this. But then: I thought to be fair I should ask three different AIs—Gemini, ChatGPT and Claude—to rank eight core specialties on a scale of one to 10 on how much they will be impacted by AI in Canada. I tend to think of AIs as being good at summarizing the conventional wisdom. But Gemini described what it did as a “qualitative synthesis based on workflow analysis and emerging health technology data.”(By the way, the visual disciplines of radiology and pathology—the AI noted—have, of course, already been the undisputed early adopters of AI.)“Amid our recent doom cycle about whether AI will lead to extreme societal harms, a parallel anxiety has taken hold in medicine about whether A.I. will one day replace physicians,” Dr. Rachael Bedard, a geriatrician wrote recently in the New York Times. But it is important to remember that AI disruption does not equal threat to demand or job loss. (The AIs anticipate those threats to be low for all specialties in Canada and especially low for the procedural specialties.)Family medicine ranks high for impact because high volume admin work is a defining burden of being an FP, and AI scribes and the AI tech to come are expected to eat into that with inbox triage, referral management and other aspects. A recent feisty LinkedIn discussion examined whether the idea of 10,000 patient primary care panels is the future of family medicine or its death? (The idea is the doctor spends most of their day with a few patients and they are supervising AI-supported workflows for the rest; but critics say it is terrible if the FP doesn’t personally know every patient.)Internal medicine: AI is great at synthesizing large amounts of lab results, imaging reports and decades of fragmented records so the impact is ranked high.Though the encounter itself won’t change for psychiatry, ambient scribes can transcribe and structure one-hour psychiatric evaluations into perfect clinical notes, and AI chatbots are seeing more and more use for patient triage and mood tracking.Of course, some of the procedural and surgical specialties have significant clinic where AI documentation will save time. And some AI help with image interpretation and planning, though the core physical work will not change. “In the short term, where AI is mostly associated with generative AI (e.g., AI scribes, CDS AI, etc.)—then, yes, AI will produce more efficiency,” Dr. Joshua Liu, co-founder and CEO of SeamlessMD, wrote to me. “But there’s already work going on with AI surgical robotics and other areas that touch the surgical realm—and in the long-term, surgical specialities will be just as impacted by AI.” We can quibble with the AI numbers but the point is: We all know AI is going to dramatically impact the daily work of physicians. And we have a system where nationally, the average gross clinical payment to a surgical specialist is about 1.72 times that of a family physician. In 2023/24, the Canadian Institute for Health Information (CIHI) reported average gross clinical payments of $556,000 for surgical specialists, $406,000 for medical specialists and $324,000 for family physicians.“There is broad agreement that physician compensation undervalues cognitive clinical work compared with procedural care,” Dr. Michelle Greiver, a Toronto FP, wrote to me. “AI-enabled efficiency should strengthen care, not become a reason to reduce fees for cognitive work. Payment models should recognize diagnostic reasoning, access, continuity, care coordination, team oversight and panel management. Although some tasks can be delegated, family physicians remain accountable for oversight and carry the clinical risk.”But it could also be argued that falling work per clinical encounter for cognitive specialties might enable some to see more patients while doing the same amount of work they did before AI—and thus narrow the income relativity gap with procedural specialists.If assistant deputy ministers in the provinces start seeing in-year upwards pressure on the primary care line as a result of AI scribes, they may look at negotiating some fees down. But the truth is—partly because of the psychology of a loss (we hate losing money probably more than we’re happy to gain some)—the doctor compensation system does not adjust easily for tech changes. (Though following the Kaplan arbitration award, Ontario introduced relativity-adjusted increases that vary by specialty. In 2025/26, psychiatry received 13.85% and ophthalmology 2.45%, compared with a 9.96% global increase.)As a longtime watcher of these issues, Dr. Eric Cadesky of Vancouver, put it to me: “The same thing that happens whenever any new technology increases procedural speed: rapid changes in income, standoffs with government and negotiated settlements (or, failing that, acrimonious unilateral clawbacks.)”A policy wonkThe Google Meet began—and there was policy wonk Will Falk, spikes of grey hair up, looking much the same as he did when I first interviewed him. At 64, he’s been retired from PwC and is moving away from commercial board work to focus on think tanks and policy activities—because the policy stuff is just so interesting now. He’s a part-time yoga instructor who lives two hours north of Toronto these days.Years ago, he negotiated on the health ministry side during Ontario physician fee negotiations and I interviewed him shortly after.The first thing we chatted about is: The current primary care research isn’t showing much in the way of AI efficiencies. But then academic studies lag current practice by 12 to 15 months and perhaps cognitive doctors have been so overwhelmed by paperwork they take AI efficiencies and get more work-life balance rather than take on more patients. But I know too many FPs who say AI scribes are saving them hours per week to not think we have seen—and will see more—AI efficiencies.I shared the question I asked Gemini about AI: “What steps could Canada’s doctors and their organizations take to speed up how quickly and fairly compensation models can adjust?”We both found it interesting that Gemini immediately said (with great confidence) that governments will move to cut per-unit fees. “In Canada they do not. Nobody has cut a nominal fee since 2012. We did it in 2011/12 on cataract and imaging, and I wore the radiology one personally for about six years afterward,” Falk said. (That is, some wouldn’t talk to him for awhile but opthalmologists were more understanding about cataracts because they had cognitive codes they wanted increased.) After that neutral question, Gemini also went right to: “The goal is to capture the value of technological efficiency for the physician, rather than letting the health ministry claw it back,” and it outlined several steps medical associations should take:establish dynamic, data-driven relativity committees bifurcate the professional versus technical components (underline the fee covers the physician’s diagnostic judgment and legal liability for the AI’s output, not just the minutes spent typing)pioneer AI-oversight billing codesaccelerate the shift to capitation and blended modelsdemand transparent ROI sharing in master agreements“No jurisdiction has designed who captures the time AI frees up. Canada funded the platforms without deciding that question. It is the one place we could lead,” said Falk.There is going to be pushback if medical associations argue some of these things. AI-oversight fee? A supervision fee stacked on an existing fee is the payer paying twice for one encounter, governments will argue. Utah’s autonomous prescribing pilot, where AI systems manage routine prescription renewals for chronic conditions, has nobody billing extra for the oversight. ROI sharing? Canada and corporations have funded the AI development so why should doctors make money from the efficiencies, governments will also argue. Falk said he would say one thing to doctors: “Don’t let anyone pay for your AI scribe because if they pay for it, they’re then going to tell you that they paid for it and that they own the productivity increase!”The futureThere’s a line in the film How to Train Your Dragon where Astrid gets frustrated with Hiccup’s distracted behaviour and says: “Our parents’ war is about to become ours!”We need young doctors to find this stuff interesting and get involved: both with their medical associations and sometimes as consultants for health ministries (to understand the physician perspective). Back during the last tech change round it involved getting granular and finding in 2011 that cataract surgery paid roughly $28 per minute while a complex hip replacement paid under $5 per minute.American health economist James Robinson famously said: “There are many mechanisms to pay physicians; some are good and some are bad. The three worst are fee-for-service, capitation and salary.”We need to use our technology—and AIs—to more effectively understand how different kinds of specialists spend their work hours. It may be that procedural specialties are most suited to activity-based funding and cognitive specialities most suited to time-based funding so we need different payment schemes for different types of specialists. Challenges with current remuneration systemsGender and other concerns: The Medical Post periodically looks at fee and income relativity in our annual money issue in articles like “The RVU solution” and “Fixing relativity: a path to progress.” This year, we follow four core procedural/surgical specialties (otolaryngology, orthopedics, general surgery and urology) and four cognitive specialties (family medicine, pediatrics, psychiatry and internal medicine). Plus dermatology and ob/gyn in the gender chart.The top chart shows—based on 2023/34 CIHI data—the gender mix of different specialties overlaid against FTE gross clinical payments.The chart below shows the ratio by province between the four cognitive and four procedural specialties. A big criticism of gross pay is that it doesn’t account for overhead. But neither CIHI, nor any government to speak of, tracks that, so overhead estimates come from self-reported surveys that often suffer from low response rates and recall bias. (A landmark study by researcher Jeremy Petch and Drs. Irfan Dhalla, Rick Glazier and Andreas Laupacis showed in 2012—using overhead estimates from the 2010 National Physician Survey—that ophthalmology drops from second to eighth net of overhead when one ranks by take-home income.)Another concern with the approach here is this blends fee-for-service and alternative payment plans into a single metric—and this doesn’t, thus not showing how provinces are trying to fix the income gap through capitation or salary models.Check out the data that made up above chart here.

AI and money: How tech could upend the doctor income relativity divide

Register for free access or login.

Related Topics