40% of Doctors Say They Should Keep the Savings When AI Makes Them Faster

More than 40 percent of doctors think they should capture most of the cost savings if artificial intelligence helps them work faster, according to the 2026 Physician Compensation Report from Doximity, reported by Healthcare Dive on August 28. That is the productivity dividend question, and healthcare is the first profession arguing about it in public. If AI makes you faster, who keeps the value, you or the organization paying you? At Metaintro, we think every worker in every sector should watch how this one resolves.

What Does the Doximity Report Actually Cover?

It is a substantial dataset rather than a quick poll. The report draws on more than 250,000 compensation survey responses collected over the last seven years, including responses from nearly 23,000 United States physicians surveyed in the past year alone. That scale matters when the findings concern attitudes, because attitude questions on small samples move around a lot.

One caveat to hold from the start. Doximity is a health technology company reporting on the market it operates in, so this is vendor research rather than independent academic work. That is not a reason to discard it, since the sample is large and the questions are specific, but it is a reason to attribute it precisely rather than treating it as neutral ground.

How Many Doctors Are Actually Using AI?

More than 65 percent of physicians use artificial intelligence on a daily or weekly basis for clinical or administrative work. That is not pilot stage adoption, it is routine use by roughly two thirds of a highly regulated profession, which is faster uptake than most sectors have managed.

The distribution is uneven in ways that matter for your own planning. Adoption is higher among non surgical specialists, which follows, since the documentation and administrative load that AI addresses is heaviest where the work is cognitive rather than procedural. Younger physicians are also more likely to use AI than older colleagues, and more likely to report that it reduces their workload.

That last point is worth separating out. Younger doctors are not only using the tools more, they are experiencing them differently, reporting genuine workload reduction. Whether that reflects better tool fluency or lower baseline resistance is not settled by the data, but it does mean the profession is not having one experience of AI, it is having at least two.

Where Exactly Are Doctors Split?

On the core question, and the split is close to even. More than a third of respondents said their compensation for a service should change if AI substantially reduces the time or effort they put into that service. However, 43 percent said compensation should remain unchanged.

Read those two together and you have the whole ethical argument in miniature. One group treats payment as compensation for time and effort, so if the effort falls, the payment should follow. The other treats payment as compensation for the outcome and the responsibility, which do not change because a tool got faster. Both are coherent positions and the profession has not chosen between them.

Where there is more agreement is on who benefits if AI increases volume rather than reducing effort. Nearly half of respondents said physicians should be the primary beneficiaries if AI enables them to complete more clinical work in the same amount of time. Only 17 percent said savings attributable to AI should translate into lower costs for patients, and 20 percent said the cost benefits should be shared.

Is the Productivity Pressure Already Arriving?

Yes, and ahead of any pay adjustment, which is the part workers in every sector should notice. A fifth of respondents say they are already facing higher expectations for productivity because of AI, and 42 percent are concerned that they will in the future.

That sequencing is the whole risk. The expectation that you will do more arrives immediately and costs the employer nothing. The argument about who keeps the value is unresolved and may never be formally settled. If those two things stay out of step, the outcome is more work at the same pay, which is how a productivity gain becomes a speedup rather than a dividend.

Healthcare is unusually exposed to this because the profession already carries a documented burnout problem, which we have covered in physician burnout and career pivots in emergency medicine and in AMA data on burnout by specialty. There is at least one encouraging counter signal, in resident doctor burnout dropping, so the trend is not uniformly bad.

Does AI Threaten Physician Jobs?

The doctors themselves do not think so, and the underlying labor data supports them. The vast majority of physicians do not see AI as a threat to their job security, and instead a meaningful share expect it to increase their compensation and career growth, with nearly a fourth saying they expect AI to boost their compensation within the next year.

Federal projections point the same way. According to the Bureau of Labor Statistics, overall employment of physicians and surgeons is projected to grow 4 percent from 2025 to 2035, about as fast as the average for all occupations, with a median annual wage of 275,930 dollars in May 2025.

So the question in medicine is not displacement, it is terms. That is a materially different problem from the one facing entry level white collar work, and it is why healthcare is the cleanest place to watch this argument play out. Nobody here is arguing about whether the job survives. They are arguing about who owns the efficiency.

Is AI Proficiency Becoming a Hiring Requirement?

It is already a factor. Thirty nine percent of doctors surveyed say AI proficiency influences hiring decisions within their specialty, and most respondents expect an earnings advantage if they stay current with AI tools compared with colleagues who do not adopt them. The report frames it as becoming increasingly important for hiring and career growth.

That is the practical finding for anyone in or entering medicine. A skill that 39 percent of a specialty already weighs in hiring is not optional in three years. And because adoption skews younger, the gap is likely to be widest exactly where experienced clinicians are competing against newer entrants who arrived fluent.

For anyone building a healthcare career, that argues for treating AI documentation and decision support tools as core clinical skills rather than as optional efficiency aids. Our guide to building a doctor profile that gets hired is worth reading with that in mind, alongside physician board certification for the credential side.

What Is Happening to Physician Pay Overall?

Slowing, which is the backdrop the AI argument sits against. Average physician pay increased by 2 percent last year, down from 3.7 percent in 2024, which the report characterizes as a sustained period of moderate growth.

Geography still matters more than the averages suggest. Physician compensation rose in 41 of the 60 metropolitan areas included in the report, with more than half of those seeing growth of at least 3 percent. So the national average conceals a real spread, and where you practise remains one of the larger levers on what you earn.

The specialty gap is the other. Surgical specialties including neurosurgery and thoracic surgery recorded the highest average annual compensation, while pediatric and primary care specialties saw the lowest, a spread we break down in our ER doctor salary guide and across healthcare occupations generally. That pattern is long standing, and it interacts with the AI finding in an uncomfortable way, since adoption is highest among non surgical specialists, meaning the doctors most likely to generate AI efficiency gains are concentrated in the lower paid specialties. We have covered adjacent pressures in physician practice bankruptcies spiking and healthcare bankruptcies across senior care and physician practices.

Why Should Workers Outside Medicine Care?

Because the same question is coming for every knowledge job, and physicians are asking it earlier and louder. Doctors have professional bodies, individually negotiated contracts, measurable output and scarcity, which is roughly the strongest bargaining position any worker holds. If the productivity dividend argument is unresolved even there, it will be harder everywhere else.

Watch which way it settles. If organizations capture the AI savings in medicine, that becomes the default across sectors, and the reasoning will be reused. If physicians secure a share, they establish a precedent that other professions can point at.

The vocabulary is worth borrowing regardless of your field. Asking who is the primary beneficiary when a tool makes you faster is a better question than asking whether AI will take your job, because for most people right now the first question is live and the second is not. The wider pattern of AI raising expectations rather than reducing load is one we examined in the coming burnout from managing AI agents.

What Would Actually Have to Change for Doctors to Share the Gains?

The mechanism is the missing piece, and it explains why the argument stays unresolved. Physician compensation is largely built on volume and on codes attached to services, so an efficiency gain does not automatically translate into anything. If a service takes less time but pays the same per instance, the doctor benefits only by doing more of them, which is the speedup outcome rather than the dividend.

Three routes exist in principle. Pay could stay fixed per service while time falls, which hands the gain to the clinician in the form of recovered hours if volume expectations hold steady. Volume could rise with compensation rising alongside it, which is the outcome nearly half of respondents said they want. Or the savings could be captured by the employer or the payer, which is what happens by default when nobody negotiates.

The complication is that the industry has not agreed how to measure any of this. Healthcare Dive notes there is still no consensus on how to assess AI's return on investment, which matters because you cannot distribute a benefit that nobody has quantified. That measurement gap is not neutral. In the absence of an agreed number, the party that already controls the contract sets the terms.

Which suggests where clinicians should apply pressure. Not on the philosophical question of who deserves the gain, but on the practical one of how it gets measured, because whoever defines the metric effectively decides the answer. Our reporting on employers and medical costs where workers disagree covers a parallel case of the same dynamic.

What Does This Mean for Your Career Right Now?

If you are a physician, get specific about your own numbers before the conversation reaches you. Know how much time your AI tools actually save, in which parts of your workflow, and what happened to your patient volume over the same period. A fifth of your colleagues already report higher productivity expectations, and the person who has measured the change is in a far better position than the person describing an impression.

Raise it at contract renewal rather than in the abstract. The report shows nearly half of physicians believe they should be the primary beneficiary of AI enabled volume gains, which means the position is mainstream within the profession rather than an outlier ask. That is worth knowing before you assume you are alone in raising it.

Treat AI proficiency as a credential. With 39 percent reporting it as a hiring factor within their specialty, and most respondents expecting an earnings advantage from staying current, this has already moved from curiosity to competitive necessity. If you are choosing where to invest limited professional development time, this is a defensible place to put it.

And if you work outside medicine, take the framing. Measure what the tools save you, and decide in advance what you think a fair split looks like, because the expectation of higher output will reach you before anyone offers to renegotiate. At Metaintro, we would rather you enter that conversation with numbers than discover the new baseline after it has already been set.

Related Articles

• High Paying Medical Careers
• How to Become a Medical Doctor
• Emergency Medicine Doctor Career Guide
• Doctor Career Categories and Specializations
• How to Become a Family Physician
• How to Become a Physician Assistant in 2026
• Minnesota Hospitals Rebound With Physician Jobs Open
• 330,000 Pharmacists and the 86,000 Doctor Shortage by 2036
• Highest Paid Nurse Practitioner Specialties in 2026
• The End of Always Available, Professional Burnout and Boundaries

People Also Asked

Q: Do doctors think AI should increase their pay?

A: They are divided. More than 40 percent say they should capture most of the cost savings when AI helps them work faster, and more than a third say compensation for a service should change if AI substantially cuts the time involved. But 43 percent say compensation should remain unchanged, so there is no professional consensus.

Q: How many physicians use AI at work?

A: More than 65 percent use it on a daily or weekly basis for clinical or administrative work, according to the Doximity report. Adoption is higher among non surgical specialists and among younger physicians, who are also more likely to say the technology reduces their workload.

Q: Is AI reducing physician workload or increasing expectations?

A: Both, depending on who you ask. Younger doctors are more likely to report reduced workload, while a fifth of all respondents say they already face higher productivity expectations because of AI and 42 percent expect to in future. The pressure is arriving faster than any agreement on how the savings get shared.

Wondering what you should earn as the work itself changes? Metaintro tracks compensation and hiring across healthcare specialties and surfaces the employers competing hardest for clinical talent. Create your free profile and price your next move against the market rather than against last year.