Short answer: both. AI will eliminate some tasks and roles, but it will also create new ones and expand others. Whether the net effect is more or fewer jobs depends on how fast firms adopt AI, how work is redesigned, and what training and policy supports are in place.
What tends to happen with big, general‑purpose tech (steam, electricity, computers)
•Near term: disruption and task displacement, especially in routine, rules‑based work.
•Medium/long term: new products, services, and business models appear; productivity gains lower costs and raise demand, which can expand employment—often in different roles than before.
Where displacement pressure is highest
•Routine cognitive/clerical: data entry, simple bookkeeping, basic customer support, templated content creation, some paralegal and analyst tasks.
•Middle-skill roles with large portions of predictable digital work.
Where augmentation and growth are likely
•Roles that mix domain expertise with judgment, creativity, or interpersonal work: product management, sales, marketing strategy, software engineering, design, research, healthcare, education, skilled trades, and operations.
•New and expanding specialties: AI operations and integration, workflow automation, data engineering/stewardship, model evaluation and safety, compliance and risk, “human‑in‑the‑loop” quality assurance, and change management.
•Sectors with in‑person or bespoke demand: eldercare, childcare, healthcare, construction, advanced manufacturing, green transition work.
What this means for workers
•Think tasks, not jobs: most roles will be partially automated; those who learn to orchestrate AI usually become more valuable.
•Build a durable skill stack:
•Domain expertise (how your industry actually creates value)
•Data/AI literacy (prompting, verification, basic statistics, automation tools, API workflows)
•Human skills (communication, negotiation, leadership, ethics)
•Make your work AI‑resilient: move toward problem framing, requirements, oversight, and client interaction; own outcomes, not just outputs.
•Show, don’t tell: keep a portfolio of AI‑augmented projects with before/after metrics (time saved, quality improved, revenue impact).
What this means for employers and policymakers
•Redesign work first, then restructure headcount; measure cycle time, error rates, customer satisfaction, and revenue/employee.
•Invest in reskilling and internal mobility; pair adoption with change management.
•Support job creation via entrepreneurship, infrastructure, and care/green sectors; strengthen safety nets to ease transitions.
Bottom line: Over the next few years, expect significant churn—some roles shrink, many evolve, and new ones appear. Individuals and firms that treat AI as a force multiplier, not a headcount lever, are most likely to come out ahead.
If you share your industry or role, I can map the most likely task changes and the best upskilling moves.