Five signs your career is harder for AI to displace

By Sophie Lindqvist

Five practical indicators frame the case for career resilience: objectively unpredictable work, active curiosity, entrepreneurship, augmentation-first employers, and skepticism toward AGI forecasts.

The analysis sets its horizon two or three decades ahead, rather than 2027 or 2028. Garrison Lovely’s Obsolete divides attitudes toward AI into three groups: “accelerators,” who expect it to solve society’s problems; “worriers,” who fear human extinction or devastating effects on work and life; and “critics,” who question claims in either direction and may regard AI as overrated. The article avoids choosing a camp, treating uncertainty as grounds neither for paralysis nor complacency.

That discipline matters because AI already performs many knowledge-economy tasks that once fell mainly to people. I, Human frames this shift as a redefinition of human identity and employability. The practical response is to strengthen capabilities AI does not reliably replace while judging future claims against what systems demonstrably do now.

What makes a job harder to automate?

• An objectively “messy” job. Luis Garicano at the London School of Economics said: “One of the best ways to resist automation is to have a chaotic, unpredictable, or ‘messy’ job.” This means work whose routine is genuinely complex, unpredictable, and nonrepetitive, not a disorganized mindset that makes orderly work look chaotic.
• AI can automate standardized tasks, but standardization comes first. Work requiring judgment across changing conditions therefore offers a stronger defense than a repeatable procedure.

What does curiosity protect?

• A “hungry mind.” Contemporary careers require work ethic, people skills, and substantial intellectual curiosity. Generative AI weakens the incentive to build that curiosity when a prompt can answer any question at once.
• The comparison is a fast-food outlet: available around the clock, quick, inexpensive, convenient, and potentially addictive, but not a substitute for sustained nourishment. The argument extends this concern to new cultural adaptations that keep mental capacity active, including “thinking retreats” for people whose thinking has become outsourced.
• Curiosity keeps workers learning and helps them understand and use AI for augmentation rather than passive consumption.

Can one person build what once required a team?

• Entrepreneurial use of AI. The value of automation lies not only in removing tedious work but in freeing time to test ideas, identify unmet needs, and connect demand with supply.
• A single person can now use AI for six functions that previously required a small staff: market research, product development, website building, marketing campaigns, customer-relationship management, and customer support.
• Professionals can likewise package expertise as new services, products, or business models that were once too expensive, time-consuming, or economically unviable to test. AI can enable work that previously looked impossible.
• Execution becomes cheaper and faster, but AI does not guarantee good ideas, sound prioritization, or an idea worth pursuing. Experimentation is not the same as thoughtless delegation.

When does augmentation protect a career?

• An employer committed to augmentation. Many organizations promise employees more interesting, fun, and rewarding work, with greater productivity and AI as a copilot or “co-intelligence.”
• Some instead automate work and cut entry-level roles to show a return on AI investment. A September 2025 Harvard Business Review article, “The Perils of Using AI to Replace Entry-Level Jobs,” addresses the danger of replacing those roles.
• Workers gain greater protection where management genuinely redesigns work to augment people rather than remove entry points. Employer and boss choice can therefore matter more than an individual’s skills, personality, or talent.

How much confidence should AGI forecasts receive?

• A forecast treated as a forecast. Lovely said in Obsolete that “much of the debate about AI’s impact on humanity rests on assumptions about the future that are presented as if they were facts.” AGI remains a prediction, not an established scientific fact.
• Defined here as machines matching or surpassing humans across nearly every cognitive domain, AGI differs from systems that perform an expanding range of intellectual tasks. Human-level reasoning, learning, and adaptation in unfamiliar situations remain a different technical proposition.
• That uncertainty neither proves AGI impossible nor removes AI’s present risks. Nor must AGI arrive for AI to displace millions of workers; its uncertainty therefore does not remove present labor-market risk.
• No one knows when, whether, or how AGI might arrive. Betting a career on the failure of technological progress would be unwise; treating Silicon Valley predictions as settled facts is no safer.

The near-term strategy is therefore practical: use AI to reclaim time for judgment, learning, experimentation, and work that changes with context. Keep developing capabilities that improve with use, and assess employers by whether they create augmented roles or merely eliminate junior positions.

AI could stop improving tomorrow, although that is unlikely, and people would still need to become more adaptable in an employment market AI has already begun to reshape. Work alongside the technology rather than wagering a career on its imagined endpoint.