Short answer: both. AI will eliminate some tasks and roles, create new ones, and change most others. In the near term you should expect more churn than collapse: many jobs will be redesigned, some will shrink, and some will grow. Whether net jobs rise or fall over the next decade depends on how quickly productivity gains translate into new products, lower prices, and higher demand.
How automation usually affects jobs
•Task, not job, automation: AI substitutes for routines (drafting, summarizing, data entry) but complements judgment, creativity, relationship-building, and hands-on work.
•Productivity → demand: When output gets cheaper/better, firms often sell more, hire for complementary work, and create new roles.
•Reallocation costs: Even if the long-run is positive, short-run displacement can be painful and concentrated.
Near-term outlook (2–5 years)
•Most exposed tasks: clerical/back-office processing, basic customer support, standard reporting, simple content generation, QA/testing, some parts of coding and analytics.
•Growing/complementary areas: AI product management, data and ML ops, model evaluation/safety, change management, compliance/governance, domain experts who can wield AI, sales and marketing amplified by personalization, designers/engineers who integrate AI into products.
•Many roles become “cyborg” jobs: same title, higher output with AI tools.
Medium-term wild cards (5–10+ years)
•If autonomous agents and robotics make big strides, more service, logistics, and some professional work could be automated.
•If progress is slower in the physical world, hands-on and hybrid human-AI roles remain resilient.
Who’s likely to benefit
•Workers who mix domain expertise with AI fluency.
•Roles requiring complex social interaction, high-stakes judgment, creativity, or work in unstructured physical environments.
•Firms that redesign workflows around AI (not just bolt it on).
Who’s at risk
•Roles dominated by predictable information processing with little customer interaction or specialized judgment.
•Organizations that freeze hiring and don’t upskill or redesign processes.
What you can do now
•Audit your tasks: list repetitive, rules-based work that AI can take; plan to redeploy time to higher-value activities.
•Build AI leverage: learn prompt design, tool orchestration, data literacy, and automation basics; document measurable productivity gains.
•Double down on human edges: communication, negotiation, leadership, cross-functional problem solving, and domain depth.
•Create artifacts: a portfolio showing how you used AI to improve speed, quality, or revenue.
•Stay adaptable: expect roles and tools to change; continuous learning beats betting on a single tool.
Bottom line: Over the next few years, AI is more likely to reshape than erase work. With smart adoption, it can create at least as many opportunities as it takes away—but the gains won’t be evenly distributed. Your best hedge is to become the person who uses AI to deliver more value, not the person doing what AI now does well.