What Jobs Will AI Replace?

Core tasks resist automation across all dimensions. 90% of work requires embodied physical care, deep human trust, and real-time clinical judgment — none of which AI can perform. Realistically 20+ years before any meaningful displacement, if ever.

Core tasks are irreducibly human — teaching young children to read, nurturing social-emotional development, safeguarding vulnerable students. 55% of work is entirely beyond AI reach, and a further 35% is augmented, not displaced. The global teacher shortage reinforces demand. 15+ years before any meaningful displacement.

Maximum Green — every signal converges. Physical work in unstructured environments, licensing barriers, surging demand, and AI infrastructure actively increasing need for electricians. AI cannot wire a building.

Near-maximum Green — every signal converges. Physical work in unstructured environments, licensing barriers, acute labour shortage, and AI infrastructure indirectly boosting demand. AI cannot fix a burst pipe behind a wall.

Core firefighting demands embodied physical presence in extreme, unpredictable environments that no AI or robot can operate in. AI augments reporting and situational awareness but cannot enter a burning building, rescue a victim, or treat a patient. Safe for 20+ years.

Core work is hands-in-mouth physical procedures that AI cannot perform. 50% of daily tasks are untouched by automation; AI augments diagnostics and admin but the dentist's hands remain irreplaceable. Safe for 20+ years.

Carpenters are among the most AI-resistant occupations — core building tasks require physical presence in unstructured environments that no AI or robotic system can replicate. Safe for 5+ years with strong wage growth and persistent labour shortages.

Manual therapy and hands-on patient care anchor this role in the Green Zone. 80% of daily work requires physical touch, clinical examination, and real-time movement guidance that no AI system can perform. Realistically 15-25+ years before meaningful displacement.

Family medicine doctors are structurally protected by licensing, personal liability, and the irreplaceable doctor-patient relationship. AI is transforming 25% of daily work — ambient documentation and care coordination — but cannot examine a patient, build trust over years, or bear legal accountability. Safe for 15+ years.

Core patrol work requires embodied physical presence, split-second moral judgment, and legal authority that AI cannot hold. AI is transforming report writing and analytics, but the officer on the street is irreplaceable. Safe for 15+ years.

Chefs and head cooks are protected by the combination of creative menu vision, palate-driven quality judgment, and kitchen leadership under pressure — tasks AI cannot execute. Back-of-house operations (scheduling, inventory, food costing) are being displaced by AI tools, but the core 65% of the role — leading people, creating dishes, and maintaining culinary standards — remains irreducibly human. Safe for 5+ years with transformation in operational workflows.

Core hands-on repair work is deeply physical and AI-resistant, but diagnostics and routine maintenance are shifting toward AI-augmented workflows. Safe for 5+ years with evolving skill demands.

Physically directing live stunts, explosions, and action sequences in unstructured environments is irreducibly human. AI previz tools transform planning workflows but cannot replace the on-set safety accountability and creative direction that define this role. Safe for 5+ years.

Augmentative and alternative communication requires hands-on device fitting, deeply personal client relationships with nonverbal individuals, and clinical judgment about communication systems that AI cannot replicate. AI improves the devices AAC Specialists configure but does not reduce demand for the specialist. Safe for 10+ years.

This role is deeply protected by irreducible cultural knowledge, unstructured physical environments, and massive government expansion — safe for 10+ years with AI augmenting monitoring tasks.

Physical installation in unstructured environments — ceiling voids, door frames, existing buildings — protects this role from automation. AI enhances analytics after installation but cannot mount readers, pull cable, or program controllers on site. Safe for 10+ years.

The hands-on, deeply interpersonal core of this role — leading reminiscence therapy, facilitating pet therapy visits, running arts and music sessions, and building trust with residents living with dementia — is irreducible by AI. Administrative and planning tasks are shifting to digital care platforms, but human presence and emotional connection IS the service. Safe for 5+ years.

The hands-on, deeply interpersonal core of this role — leading activities, building relationships with residents, and combating isolation in care settings — is irreducible by AI. Administrative and planning tasks are shifting to digital tools, but the human presence IS the service. Safe for 5+ years.

The actuarial profession's extreme credentialing barrier (FSA/FCAS — 7-10 exams over 5-7 years) and regulatory mandate for human sign-off create a durable moat. AI is automating the computational core but the actuary's judgment, accountability, and certification role is irreplaceable. Safe for 5+ years; the role transforms from model builder to model governor.

Acupuncture is a hands-on, body-contact healthcare practice where needling technique, tactile diagnosis, and the therapeutic relationship are inseparable from the service itself. Safe for 10+ years with no viable AI or robotic substitute for clinical practice.

This role is structurally protected by physician licensing, bedside physicality, and irreducible moral judgment — but AI is reshaping how documentation, diagnostics, and treatment planning are done. Safe for 5+ years; daily work evolving now.

This role is structurally protected by licensing, liability, and the irreducible therapeutic alliance that drives addiction recovery. Documentation is shifting to AI, but the clinical core — MAT prescribing, dual diagnosis management, crisis intervention — remains firmly human. Safe for 5+ years.

Due process mandates, personal accountability for binding decisions, and deep structural barriers protect ALJs from AI displacement. AI transforms research and case administration but cannot preside, evaluate credibility, or issue legally binding determinations. Safe for 10+ years.

Assessing prospective adopters, matching children with permanent families, and navigating the legal and emotional complexities of adoption are irreducibly human tasks — requiring trust built over months, moral judgment about lifelong family formation, and professional accountability for irreversible placement decisions. AI is reshaping documentation and case administration, but licensing barriers, personal liability, and profound cultural resistance to algorithmic adoption decisions protect this role. Safe for 7+ years, with daily workflow transformation.