Short answer: both. AI will eliminate some jobs, change most jobs, and create new ones. Whether it’s a net gain or loss depends on how fast firms adopt AI, how demand responds to lower costs, and what governments and workers do to adapt.
What to expect
•Near term (next 3–5 years): More displacement than creation in certain white‑collar, routine cognitive roles. Many other roles get “AI copilots” that raise productivity rather than replace people. Global estimates suggest modest net job losses in the short run, with wide variation by country and sector.
•Medium to long term: Historically, general‑purpose technologies create more jobs than they destroy by spawning new products, services, and industries. That’s likely here too—but the gains are uneven and require reskilling and policy support.
Most exposed tasks and roles
•Routine cognitive/administrative: data entry, basic bookkeeping, scheduling, standard reports, form processing, basic customer support, simple coding, first‑draft content.
•Parts of legal, accounting, and research involving document review and synthesis.
•Some marketing production work (ad copy variants, simple design).
Roles likely to grow or be reshaped
•Augmented professionals: software engineers, designers, analysts, clinicians, lawyers, educators—doing higher‑value work as lower‑level tasks automate.
•“AI supply chain”: data engineering, ML engineering, MLOps, evaluation/red‑teaming, AI safety/governance, compliance, cybersecurity, privacy.
•Human-in-the-loop and integration: workflow designers, prompt/tooling specialists inside functions (finance, ops, sales), change management.
•Services hard to automate: skilled trades (electricians, plumbers, HVAC), healthcare and eldercare, early education, hospitality, field technicians.
•New product/service categories we can’t fully foresee yet, enabled by cheaper intelligence and automation.
Why both creation and destruction happen
•AI automates tasks, not whole jobs; most jobs rebundle into higher‑value tasks.
•Productivity gains can lower costs, increase demand, and expand employment—if firms reinvest and markets grow.
•Frictions (retraining time, regulation, trust, liability) slow full automation, stretching the transition.
How to future‑proof yourself
•Blend domain expertise with AI fluency: learn to use AI to research, draft, code, analyze, and automate your routine tasks.
•Lean into human advantages: client relationships, negotiation, leadership, problem framing, cross‑functional coordination, ethics/judgment.
•Move up the value chain: from doing tasks to designing systems and making decisions.
•Build data and automation literacy: basic scripting, workflow tools, API usage, data hygiene.
•If you work in at‑risk roles, pivot toward:
•Roles that deploy AI inside your function (ops analytics, RevOps, FP&A with AI, QA/evaluation).
•Customer‑facing and revenue‑driving work (sales, account management).
•Physical‑world services with complex context or trust requirements.
Policy and employer choices that matter
•Broad access to reskilling, portable benefits, and support during transitions.
•Encouraging new firm formation and diffusion of AI beyond big companies.
•Clear rules on accountability, safety, and data use to speed responsible adoption.
Bottom line: In the short run, expect visible displacement in routine cognitive work and widespread job redesign. Over time, with adoption, retraining, and growing markets, AI is likely to be a net job creator—but the benefits won’t be automatic or evenly distributed. If you share your role and tasks, I can assess your exposure and suggest concrete upskilling steps.