TD Economics: AI Reshapes Jobs Rather Than Eliminating Them

Only 1.8% of U.S. jobs today could have more than half of their regular tasks fully automated with current AI and software tools. That single figure, cited in new TD Economics research, frames why the report's authors expect AI to change how Americans work rather than eliminate large numbers of them.

The analysis comes from Rannella Billy-Ochieng', Senior Economist, and Thomas Feltmate, Director and Senior Economist at TD Economics. Their verdict is calm: "There will be a gradual change, rather than a rapid and significant disruption," they wrote.

How widely is AI actually used?

A 2026 TD AI Insights survey of 2,500 Americans found 83% use AI-powered tools to support their work. Yet comfort hasn't translated into fear of replacement: 68% of respondents said they were less worried about AI fully taking over their job than they were a year ago, and 91% believed AI should be used in the workplace.

"The future of work isn't about humans versus AI," said Ted Paris, Head of Analytics, Intelligence & AI at TD Bank U.S. "It's about how people and technology work together to unlock productivity, innovation and better outcomes."

Even so, respondents stressed that human involvement and judgment remain critical. AI can recommend; people still have to check.

Why a 'job apocalypse' isn't the base case

The economists argue large-scale AI-driven job losses would need several conditions at once: AI tools that accurately and completely automate decisions, firms able to fund those systems and clear operational hurdles, and rapid full adoption. Each condition alone is a bottleneck.

The mechanism matters here. Most jobs are bundles of tasks. Automating one task doesn't remove the person doing the others — which is why that 1.8% figure is so small. Fully autonomous AI remains rare, even as today's leading models advance in areas like software coding.

The economics don't add up — yet

Ability isn't affordability. An MIT study on computer vision systems found 36% of occupations include at least one task that could technically be automated — but only 8% include a task that's economical to automate at scale. Beyond the technology itself, companies must pay for implementation, employee training, compliance, cybersecurity and ongoing oversight, and they face uncertainty over future costs of computing power and specialized hardware.

Savings are real for routine work like summarizing documents, drafting emails and extracting information. Costs climb sharply once tasks demand complex reasoning, planning or decision-making.

Where the pressure is showing up

Business adoption has lagged consumer uptake, with many firms still experimenting, and day-to-day use varies by industry. That slower pace gives workers and employers time to adjust.

The adjustment is already visible — not in mass layoffs but in hiring. According to the report, many employers are scaling down hiring rather than cutting staff, and the pinch lands mainly on younger workers, who are finding fewer entry-level openings. Meanwhile, growth in AI infrastructure, data centers and related industries is generating new jobs.

"The ultimate employment impact depends not just on the jobs displaced, but also on the economy's ability to create and absorb workers into new ones," Billy-Ochieng' and Feltmate noted.

For now, the practical read for any worker is straightforward: AI is most likely to sit beside you as a digital assistant — summarizing, drafting, organizing — while the judgment, creativity and relationship-building parts of your job stay yours.