Short answer: both—AI will eliminate some jobs, reshape many more, and create new ones. Whether the net is positive or negative depends less on the technology itself and more on how quickly businesses redirect productivity gains into new products, services, and markets.
What will decide the balance
•Speed of adoption vs. time to retrain: Rapid rollout without upskilling drives displacement; phased adoption with training shifts tasks rather than jobs.
•Demand elasticity: If AI lowers costs and prices, and demand expands (as it often does), firms hire more—even as tasks automate.
•Complementary bottlenecks: Jobs that bundle AI with human judgment, relationships, domain knowledge, or hands-on work in messy real-world settings are more likely to grow.
•Policy and management choices: Investment in worker training, mobility, and innovation tilts outcomes toward job creation.
Where displacement risk is highest
•Routine cognitive and back-office tasks: data entry, basic bookkeeping, standard customer support, rote research/summarization, low-complexity reporting.
•The “middle” of some professions: parts of legal review, basic coding, radiology reads, or marketing production—when work is standardized and quality is easy to verify.
Where opportunities are growing
•AI-augmented professional work: software engineering, design, marketing, sales, finance/FP&A, and healthcare—where humans set goals, verify outputs, and handle exceptions.
•New and expanding roles: AI product and workflow designers, data/ML ops, AI safety/governance/compliance, evaluators and auditors, cybersecurity, human-in-the-loop QA, robotics techs.
•Sectors with secular tailwinds: healthcare and elder care, green/energy transition, advanced manufacturing, infrastructure, education and training, and security/public safety.
What individuals can do now
•Unbundle your job into tasks. Tag each as automate, augment, or uniquely human. Aim to spend more time on the last two.
•Get fluent with AI tools in your field. Treat them like power tools: write clear instructions, build checklists, and always verify.
•Double down on moats AI doesn’t replace: domain expertise, communication, collaboration, ethics/compliance responsibility, and customer trust.
•Make your value visible: track before/after productivity, cost, quality, and speed using AI; use that to move up to higher-scope work.
•Keep a learning cadence: one new tool per quarter, one portfolio project per quarter that shows AI-augmented results.
For managers and policymakers
•Redeploy productivity gains into growth (new offerings/markets) rather than pure headcount cuts to drive net job creation.
•Fund structured upskilling and internal mobility; measure outcomes, not seat time.
•Build governance: data quality, evaluation, security, and human-in-the-loop checkpoints.
Bottom line: Over a 5–10 year horizon, history and current adoption patterns suggest net job growth is plausible, but with significant churn and painful transitions in specific roles. The best predictor of outcomes for any one person or company is how deliberately they pair AI with human strengths and invest in new demand.
If you share your industry or role, I can map likely task shifts and suggest concrete upskilling steps.