Ask HN: Which skills are least likely to be replaced by AI?

Overall views on replaceability and timelines

• Some argue that, given enough time, essentially all skills can be automated; the question is “when,” not “if.”
• Others stress that near-term limits in robotics, perception, and economics mean many jobs are safe for decades.
• Several comments warn against sci‑fi extrapolation and “magical thinking” about superhuman androids when current systems still fail at basic tasks like general driving.

Physical and manual trades

• Strong consensus that jobs requiring fine motor skills in messy, cramped, variable environments are last to go: plumbing, HVAC, electricians, carpenters, roofers, locksmiths, nurses, physios.
• Counterpoint: specialized robots (e.g., roofing bots) are already emerging; skeptics of “trade safety” compare this to past underestimation of AI in Go and art.
• Many note that by the time androids can do full plumbing, nearly all work will be automated anyway.

Emotional, interpersonal, and charisma-based work

• Some believe in‑person charisma, genuine human contact, and care work (nursing, childcare, psychotherapy, teaching-as-socialization) are highly resilient.
• Others argue robots/avatars can simulate emotions consistently and may outperform average humans in many service roles; debate over whether charisma itself can be “faked” convincingly.
• Rich clients may keep paying for human therapists; mass-market mental health could shift to LLM-based systems.

Art and creativity

• One camp claims “real art” as self‑expression is inherently human and non‑replaceable.
• Another notes that most commercial art (logos, concept art, UI, etc.) is already heavily disrupted by generative models.
• Disagreement on current AI output: some see it as formulaic and shallow; others find it practically useful, especially when humans act as art directors.

Investigation, science, and critical thinking

• Some assert investigative work (detectives, scientists) is uniquely human due to open‑ended hypothesis formation and active evidence gathering.
• Others counter that LLMs already propose theories and protocols, and see no principled barrier to more.

Responsibility, law, and regulation

• Roles like doctors, pilots, captains, and lawyers may persist because someone must be formally responsible and licensed, even if AI does most of the work.
• Legal work is seen as highly automatable at the document‑review and drafting level, but professional gatekeeping and regulation may slow full replacement.

Economic and societal framing / career strategy

• Automation is constrained by economic feasibility, unions, and regulation, not just technical possibility.
• Some foresee “Certified Human-Produced” niches and elites preserving certain human roles while automating others.
• Several suggest embracing AI‑augmented fields rather than trying to hide from AI, though others argue essential trades (e.g., sewage repair) will remain valuable regardless of AI-driven prosperity elsewhere.