Short answer: Both. AI will eliminate some jobs, create others, and change most. Over the next 5–10 years, expect sizable task-level automation and job churn; whether net jobs rise or fall will depend less on the tech itself and more on how fast organizations redesign work, create new products, and retrain people.
What determines whether AI is net positive or negative for jobs
•Task mix: AI automates tasks, not whole occupations. Jobs with many routine, rules-based tasks see more substitution; jobs rich in interpersonal, creative, or physical-world tasks see more complementarity.
•Demand response: When AI cuts costs or boosts quality, lower prices and new features can expand demand enough to offset automation losses (e.g., more customer support offered because it’s cheaper).
•New-task creation: Major technologies historically spawn new roles, firms, and industries. The speed of that creation vs. automation matters.
•Reorganization lag: Firms need time to redesign workflows and products to fully use AI. During this lag, displacement can outpace hiring.
•Policy and mobility: Training, safety nets, labor-market flexibility, and entrepreneurship shape how smoothly workers shift to growing areas.
Where displacement risk is higher
•Routine cognitive and back-office work: data entry, basic bookkeeping, standard reporting, some paralegal and compliance tasks.
•Template-based content and support: first-line customer service, basic marketing copy, FAQs, form emails.
•Some programming and analytics tasks: boilerplate code, QA, routine SQL, documentation.
Where resilience or growth is more likely
•Roles needing human presence or dexterity in messy environments: skilled trades, field technicians, logistics, some manufacturing, hospitality.
•Trust- and relationship-heavy work: sales, account management, counseling, nursing, education, management.
•High-end knowledge work that uses AI as leverage: product management, data science, R&D, design, law and consulting for novel/complex matters.
•New AI-adjacent roles: workflow designers, AI product leads, data stewards, evaluators/auditors, safety/ethics, prompt/tooling specialists, change managers.
What we’re already seeing in studies and pilots
•Large productivity gains on specific tasks (often 20–40% faster) and bigger boosts for less-experienced workers when given AI copilots.
•Quality improves on well-structured, common problems; performance can drop on novel, high-stakes, or edge cases without oversight (“jagged frontier”).
•Companies that pair AI with process redesign capture far more value than those that just “bolt it on.”
Practical takeaways
For individuals
•Become “AI-augmented” in your domain: learn to decompose your work into tasks, then offload drafting, summarizing, retrieval, data cleanup, and routine code.
•Double down on human advantages: problem framing, judgment, persuasion, collaboration, and field execution.
•Build T-shaped skills: a strong domain plus data/AI literacy (prompting, verification, basic scripting, workflow automation).
•Keep evidence of impact: track output/quality gains from AI use to strengthen your bargaining power.
•Hedge against disruption: broaden your task mix, cultivate client-facing skills, and stay employable across adjacent roles.
For companies
•Redesign work, don’t just deploy tools: map tasks, automate the right ones, and redefine roles around what humans do best.
•Invest in enablement: targeted training, internal “model marketplaces,” data quality, and governance.
•Redeploy rather than just reduce: move people from automated tasks to revenue-generating, safety, or quality work.
•Measure outcomes: productivity, error rates, customer satisfaction, and time-to-market—not just license counts.
•Communicate and co-create: involve frontline staff in identifying useful automations to raise adoption and trust.
For policymakers/educators
•Speed up skill transitions: short, stackable credentials tied to employer demand; apprenticeships for AI-augmented roles.
•Modernize safety nets and mobility supports: job search, relocation help, childcare, portable benefits.
•Encourage new-firm formation and diffusion to turn productivity into job creation.
Bottom line: AI won’t “take all the jobs,” but it will reshape nearly all of them. In the near term, expect some net displacement in routine knowledge work; over the longer term, net job growth is plausible if businesses and policymakers convert productivity into new products, services, and industries—and if workers actively upskill to make AI their collaborator rather than their competitor.