AI in Spanish Healthcare: The Debate on Job Displacement

AI vs Healthcare: Who Survives the Adjustment?

A dentist ending up cleaning teeth. A surgeon replaced by a robot that doesn't tire, get depressed, or charge for on-call shifts. The thesis circulating in the debate is that artificial intelligence won't just reach factories: it will enter the consultation room, the operating theater, and the civil servant's counter. And professionals who have spent decades inflating prices and arrogance will discover their degree offers no protection. The discussion, with 139 accumulated replies over nearly two years, has shifted from mockery to nuance: some see an imminent revolution, while others recall we've been hearing the same thing for a century.

Machines Already Handle Diagnosis

The most repeated argument is that AI already outperforms the average doctor in what they do most of the day: reading records, cross-referencing symptoms, interpreting tests. A bot has access to the entire file, published scientific literature, and the patient's family health data; it doesn't tire, get distracted, or have bad days like anyone else. Anyone who has suffered at the hands of a bad doctor knows what this means.

The response from the other side is that diagnosis is the easy part. The hard part is everything else. Detecting when a patient lies about taking drops, smelling alcohol on breath at ten in the morning, noticing someone says they feel fine while their vvife looks on with concern. That doesn't show up in any blood test. Illness, it is argued, is rarely black or white; it is almost always a spectrum of grays requiring a human presence.

How Long Until AI Replaces Doctors?

Here the debate splits. One camp argues that it won't even take five years before most white-collar jobs, including healthcare roles, are under threat. The other reminds us we've heard for a decade that trucks and buses would drive themselves, yet they still need drivers. That virtual reality was going to change everything but remained anecdotal. That miracle batteries have been announced for years but haven't arrived.

The explanation given for this slowness isn't technical, but economic and political: most jobs don't exist due to production needs, but as excuses to distribute income. Notaries and registrars could have disappeared a century ago with a public registry system, yet they remain, charging for procedures a computer system would solve in seconds. If those haven't fallen, why would the consultation?

The State Sustaining the Bubble

The initial message points directly at the public sector: the State protects the inefficient because every public job is a vote. Criticism extends to civil servants in general, with a circulating calculation regarding each group's weight within public employment: education and university would account for more than a quarter of the total, healthcare around a fifth, and justice, security, and defense together would fall below that. University, it is added, dedicates one in three positions to administration and services.

Against this, the defense of public employment doesn't discuss efficiency: it discusses what happens to people. If AI can replace a third of the staff, where does the rest go? The pragmatic answer given is that it's better to keep them employed, paying income tax (IRPF) and consuming, than swelling unemployment lists in an economy that, it is argued, produces nothing and lives on an internal circuit of public money.

Legal Responsibility Will Always Be Human

One of the strongest arguments against total replacement isn't technical or economic: it's legal. You can't sue a robot. There will always be a need for a legally responsible party, and that party will be a person. The manufacturing company won't assume the fruta and financial risk of a system failure when it can offload it onto a cheap, replaceable human supervisor.

The counter-argument is that this human supervisor is a decorative figure. Putting someone to watch a system they don't understand is like putting a monkey to supervise work it doesn't comprehend. Real supervision, it is argued, will come through code audits and cameras, via companies independent of the bot owner, not with a person sitting in a chair staring at a screen.

Diagnostic Costs and Fine Print

The promise of cheaper, more accurate healthcare clashes with a detail almost nobody mentions: the law of diminishing returns also applies to diagnosis. They won't give you a full-body MRI because your ear itches. With AI, it is argued, fewer tests will be done, only when diagnostic yield is high, purely for economic efficiency.

And there appears the uncomfortable scenario: in private healthcare, AI will say there is a one percent probability of a serious illness, that necessary tests are excluded from the policy for not reaching the established threshold, and that the patient decides whether to pay out of pocket. Technology lowers diagnostic costs, but not necessarily access.

---

The scenario drawn by the optimists is seductive: the ambulance arrives sooner, the accident victim goes straight to the operating room, surgery lasts minutes, and recovery is monitored without a single human error. No fatigue, stress, depression, or indifference. Compared to what exists, the leap seems evident. The unanswered question remains: who pays the bill for that system, who assumes liability when failures occur, and what happens to the millions of people whose job consists precisely of being the redundant human?