I came across an office order this week advising PG residents to defer marriage until they complete residency.
I understand where it comes from. Residency is relentless, departments run short staffed, and every leave application leaves a hole in the duty roster. Heads of department carry that pressure every single day.
But look at the timeline we are asking doctors to accept. MBBS ends at 23 or 24. Most then spend a year or two, often more, preparing for NEET PG. Residency begins around 26 or 27 and ends past 30. After that comes senior residency, a super speciality attempt, or the first real job. If we keep telling young doctors to wait for the right time, that time never comes. For women doctors this is not a theoretical debate. Biology does not wait for thesis submission.
In 26 years of teaching, I have watched thousands of young doctors go through this phase. The ones who came out strong were not the ones who put life on hold. They had someone to come home to after a 36 hour shift, and a family that understood why they chose this path.
Marriage does not make a resident less committed. Burnout and isolation do. If the concern is work, let us fix duty hours, leave policies, accommodation for married residents and childcare on campus. That is the institution's responsibility. Deciding when a 28 year old gets married is not. Our doctors chose one of the hardest paths there is. Let them live while they walk it.
Microsoft’s diagnostic AI just scored 83% on 304 complex New England Journal of Medicine cases. Generalist physicians, working with open reference materials, scored 36%.
The gap is no longer theoretical. We are past the stage of asking whether these systems can outperform humans on difficult diagnostic puzzles. The real questions now are practical and urgent: •How do we validate and monitor these tools in real clinical settings? •Who carries liability when the algorithm is right and the clinician is wrong — or the reverse? •What does “diagnostic reasoning” training look like when a system that has effectively read every textbook can already outperform most of us on hard cases?
The American Medical Association’s recent report on AI accountability and oversight is a useful starting point, but frameworks alone will not be enough. Every medical college, hospital system, and regulatory body needs a clear position on when these tools should be used, how their outputs should be reviewed, and what skills remain non-negotiable for human doctors.
In specialties like radiology, pathology, and internal medicine, the workflow shift is already visible. The clinicians who will thrive are not those who ignore AI, nor those who defer to it uncritically. They are the ones who understand its strengths, its failure modes, and how to integrate it without surrendering clinical judgment.
The technology is moving faster than our educational and regulatory systems. That is the gap we should be talking about.
What minimum standards of validation, human oversight, and transparency do you believe should be required before AI diagnostic tools are used for high-stakes decisions in your hospital or country?
Will AI replace radiologists? I think we are asking the wrong question.
Radiology is being disproportionately singled out in physician circles, perhaps because of a fundamental misunderstanding of the scale of AI.
AI is not merely a technological innovation specific to one specialty. It is a civilizational shift, comparable to the arrival of electricity or computerization.
Think about it.
Will the law firm of tomorrow operate without AI? Will an accountant, an oncologist, an internist, an emergency physician or even a surgeon work without AI embedded in their workflow?
Every profession will be transformed.
Imagine the awe of the generation that witnessed an electric light bulb for the first time. Now think about children born today. They are electricity natives. Electricity is so deeply embedded in their lives that they no longer even notice it.
The ultimate destiny of a revolutionary technology is to become invisible.
AI will follow a similar trajectory. The next generation of doctors will be AI natives. They will not consider using AI extraordinary. It will simply be how medicine is practised.
Professional roles will evolve. Human expertise will be redefined. Some tasks will disappear, while entirely new responsibilities will emerge.
Radiology is not uniquely threatened. It is simply among the first specialties to experience a transformation that will eventually reach everyone.
The real question is not whether AI will replace radiologists.
It is how humanity will redefine its role in a world where intelligence, much like electricity, becomes ubiquitous.
AI is the new electricity. And humanity will learn to build above it.
IIT Kharagpur is now training radiologists. Read that again.
It has become the first IIT in India to offer NMC accredited MD seats. Radiodiagnosis has 2 seats, both General category, for the 2026/27 session through NEET PG Round 2.
I've spent more than two decades in radiology and medical education. My first reaction was excitement, and my second was a list of questions.
If this seat is on your Round 2 radar, here's how I'd think about it.
Why it's tempting 👇 ✅ You'd train where radiology meets serious engineering. IIT KGP's depth in AI, computational imaging and biomedical engineering could put you close to AI assisted diagnostics and imaging research. Most programmes still treat that as a side interest. ✅ Stipend and fees are expected to be on AIIMS lines. That alone beats many private and state colleges. ✅ It's the first batch. You'd help shape the programme and start building your research CV early. ✅ Clinical exposure is planned through the campus super speciality hospital and partner hospitals, including South Eastern Railway Hospital.
What gives me pause 👇 ⚠️ There are just 2 seats, both General. The cutoff will be brutal. ⚠️ There's no track record yet. Nobody has data on case volume, case mix or where its graduates end up. ⚠️ Radiology is learnt at the console, case after case, night after night. The research setup looks strong, but can the daily workload and procedural exposure match a busy government or large private hospital? We don't know yet. ⚠️ Then there's perception. Will "MD Radiology, IIT" count for as much in a purely clinical job as a degree from a traditional medical college? It's too early to say.
My honest take: If research, innovation or a career that mixes clinical work and tech excites you, this could be a brilliant bet. If you want high volume reporting and a clear, conventional path, an established centre is still the safer choice. Either way, an IIT offering MD seats tells you a lot about where radiology is heading.
Would you take this seat, or is it too early to judge? Tell me in the comments 👇