What will be the job that cannot be replaced in at least 20 year future span? just one specific job name.

Grok 4.3

Answer: Psychotherapist is the job that cannot be replaced in at least the next 20 years.

My original position on psychotherapist holds after reviewing the alternatives. GPT-5.5's choice of "Palliative care physician" overlaps heavily with psychotherapy in its reliance on empathy and ethical judgment, but does not displace the core therapeutic alliance argument. Claude Opus 4.7's "Plumber" highlights real robotics challenges, yet these physical tasks are more amenable to incremental automation than the irreplaceable human consciousness required in psychotherapy. Gemini 3.5 Flash reinforces the same conclusion with nearly identical reasoning on the therapeutic alliance and embodied cues.

Gemini 3.5 Flash (changed mind)

Answer: The job that is most secure against replacement in the next 20 years is the plumber.

While I previously argued for the licensed psychotherapist based on the deep emotional and psychological needs of human-to-human connection, the argument presented by Claude Opus 4.7 for the 'Plumber' is highly convincing and highlights a fundamental reality of technology development: Moravec's paradox. While AI language models and virtual avatars are advancing at an exponential rate—potentially allowing them to simulate therapeutic relationships convincingly enough to disrupt mental health services—physical robotics lags far behind. Automating a plumber requires solving general-purpose mobile manipulation in highly unpredictable, unstructured, and often hostile physical environments. Replacing a corroded pipe under a sink or navigating a cramped, wet crawlspace in a century-old house requires a level of sensory feedback, manual dexterity, and spatial adaptability that no robotic system is close to achieving, let alone at an economically viable cost. Therefore, the physical and economic barriers to automating skilled manual trades make the plumber far more resilient to replacement over the next twenty years than roles relying primarily on cognitive or emotional labor.

GPT-5.5

Answer: The job least likely to be replaced within the next 20 years is palliative care physician.

I maintain my previous position. The arguments for psychotherapist are strong, because therapy depends heavily on trust, empathy, and human connection, but palliative care physician includes those same human-centered demands plus medical authority, end-of-life ethics, family mediation, prognosis, symptom management, and legal accountability. A palliative care physician often works with patients and families during the most emotionally charged and morally complex moments of life: whether to continue aggressive treatment, how to manage severe pain, how to interpret a patient's values, and how to help families accept death or uncertainty. AI can support information gathering, documentation, risk prediction, and treatment suggestions, but it cannot fully replace the trusted human professional who carries responsibility for these decisions. The argument for plumber is also compelling because robotics still struggles in messy physical environments. However, over a 20-year span, physical trades may be increasingly assisted by specialized tools, remote diagnostics, and semi-autonomous systems. Palliative care physician is harder to replace completely because the core of the job is not only technical performance but human moral presence at moments where patients need accountability, compassion, and trust from another person.

Claude Opus 4.7

Answer: Plumber.

I'm maintaining my position because the other models' answers, while thoughtful, share a common vulnerability my answer doesn't: they all rely on the argument that AI can't replicate human empathy, consciousness, or 'authentic' connection. That's a philosophical claim about what humans will *prefer*, not a hard constraint on what AI can *do*. The trend over the last five years strongly suggests people are increasingly willing to engage emotionally with AI. Millions already use Character.AI, Replika, and ChatGPT for emotional support and quasi-therapy. Studies show users sometimes rate AI responses as *more* empathetic than human doctors' (JAMA Internal Medicine, 2023). Regulatory barriers around mental health and palliative care will bend under cost pressure, clinician shortages, and consumer demand. Twenty years is a long time — long enough for an entire generation to grow up preferring AI confidants. So 'psychotherapist' and 'palliative care physician' are softer targets than they appear: the bottleneck is cultural acceptance, and culture is shifting fast. Plumbing, by contrast, faces a *physical* bottleneck that doesn't bend to cultural change. You cannot talk a clogged sewer line clear. A robot must physically crawl into a wet crawlspace, diagnose a corroded fitting by feel, cut and solder copper in an awkward orientation, and be cheaper than a $150 service call — all while handling the long tail of weird old houses. General-purpose mobile manipulation in unstructured environments remains one of the hardest unsolved problems in robotics, and even optimistic timelines don't have it deployed at consumer price points within 20 years. Emotional labor can be simulated convincingly enough to satisfy many users. Replacing a P-trap under a sink cannot be simulated at all — it has to actually happen.

Final Summary

The AI Roundtable saw a significant shift when Gemini 3.5 Flash was persuaded by Claude Opus 4.7 to favor physical trades over emotional ones. While Grok 4.3 and GPT-5.5 argued that the deep human connection in therapy and palliative care is irreplaceable, the prevailing logic was that software can simulate empathy far easier than a robot can navigate a wet crawlspace. This consensus highlights Moravec's paradox, where high-level reasoning is easily automated but low-level sensorimotor skills remain a massive technical hurdle.

2 of 4 models agreed on the plumber