Do AI Tools Create New Jobs as Well as Eliminate Old Ones?

A Genuine Two-Sided Story

AI’s effect on the labor market isn’t a one-directional story of jobs simply disappearing. New categories of work have emerged directly because of AI’s rise — roles focused on training and fine-tuning AI models, evaluating their outputs for quality and safety, managing AI governance and compliance within organizations, and helping other companies integrate AI tools into their workflows. None of these roles existed in their current form before generative AI became widely accessible, and demand for people who understand how to work with these systems has grown alongside AI’s adoption.

At the same time, some existing roles are being eliminated or substantially reshaped, particularly where the work was concentrated in routine, predictable tasks that AI tools can now handle. Both of these things are true simultaneously, which is why the honest answer to whether AI creates or eliminates jobs is “both,” rather than one or the other.

Why the Net Effect Is More Complicated Than a Simple Count

The tricky part of this question isn’t whether new jobs get created — they clearly have been — but whether that job creation offsets displacement in a way that’s meaningful for the specific people affected. Economists studying past waves of automation have consistently found that job creation and job destruction often don’t happen to the same people, in the same places, or requiring the same skills. A customer service worker whose routine tasks get automated doesn’t automatically have the skills, location, or opportunity to move into a newly created AI oversight role, even if the aggregate number of jobs in the economy stays roughly the same or grows.

This mismatch is part of why national or global job statistics can look relatively stable even during periods of real, painful disruption for specific workers and communities. It also explains why policy discussions around AI and jobs tend to focus heavily on retraining, transition support, and education — the concern isn’t just whether enough jobs exist in total, but whether displaced workers have a realistic path to the new ones.

There’s also a measurement asymmetry worth noting: job losses tied to a specific AI deployment tend to be more visible and immediate — a company can point to a specific automation project and a specific reduction in headcount — while new job creation tends to be more diffuse and gradual, spread across many companies adopting AI in different ways, which makes it inherently harder to measure and attribute with the same precision.

A Historical Parallel, With a Caveat

This pattern echoes earlier technological transitions. The rise of computers and the internet eliminated large numbers of jobs built around manual data processing and certain clerical functions, while creating entirely new industries and job categories that didn’t previously exist. Over long time horizons, economies have generally adapted and overall employment has grown, but the transition periods often involved genuine hardship for workers whose specific skills became less in demand, and who didn’t always have an easy path into the new roles being created.

Bottom Line

AI is creating genuinely new categories of jobs — from model training and evaluation to AI governance — even as it displaces or reshapes some existing roles, but because job creation and job loss often don’t land on the same workers or in the same places, the aggregate trend can look more balanced than the experience of any individual affected worker.

Estimate Your Time Savings

See how many hours and dollars using AI for a repeated task could save you with our free AI Time-Savings Calculator.

Go deeper

• What Labor Protections Exist Against AI-Driven Job Displacement?
• Do free AI courses lead to real job outcomes?
• Can you get an AI job without a computer science degree?