Short answer: both. AI will automate parts of many jobs, fully replace some roles, and create new ones. Most evidence points to heavy job “churn” in the next 5–10 years, with a small net loss or near balance in the short term and potential net gains longer term if productivity translates into new demand and if workers can retrain.
What tends to be automated first
•Routine, repeatable cognitive tasks: data entry, basic bookkeeping, scheduling, summarization, first-draft writing, standard customer support, some QA/testing.
•Parts of knowledge work: drafting code, generating reports, contract review triage, simple research.
What tends to grow
•Roles that build, deploy, or govern AI: AI/ML engineering, data engineering, MLOps, AI product management, AI security, risk/compliance, auditing, model evaluation, synthetic data, tooling.
•Hybrid roles: domain experts using AI (healthcare, law, finance, marketing, design, education).
•Human-centric services hard to automate end-to-end: healthcare, skilled trades, field work, sales, management, caregiving, hospitality.
•New business creation enabled by lower startup and operating costs.
What major studies suggest
•World Economic Forum (2023): large churn by 2027; tens of millions of jobs both created and displaced, with a small net decline in the near term.
•OECD (2023): a sizable share of jobs are highly exposed to AI; outcomes depend strongly on adoption, worker training, and job redesign.
•McKinsey (2023–24): big occupational shifts this decade (e.g., declines in office support, increases in healthcare, STEM, and creative/management roles); generative AI could boost productivity and, if coupled with demand, employment.
Key dynamics to understand
•AI usually automates tasks, not entire occupations; many jobs will be redesigned rather than eliminated.
•Complementarity matters: workers who pair AI with domain expertise, judgment, and interpersonal skill tend to become more valuable.
•Transition speed is the risk: even if long-run employment is stable or higher, short-run displacement can be painful and unequal (middle-skill clerical roles are particularly at risk).
How to future-proof yourself
•Become AI-literate: learn to use AI tools deeply in your field; automate parts of your own workflow.
•Double down on human strengths: problem framing, ethics, leadership, teamwork, negotiation, empathy, hands-on work.
•Build durable moats: domain expertise, data literacy, systems thinking, change management.
•Keep a portfolio of evidence (projects, automations, outcomes) and be willing to reskill or switch roles as industries shift.
Bottom line: AI will both create and take away jobs. In the near term, expect more disruption than net creation in some sectors; over the longer run, outcomes depend on how effectively businesses, workers, and policymakers convert productivity gains into new products, services, and inclusive growth.