AI Is Already Taking Work Away From Coworkers

One in five U.S. workers now says AI handles at least one task previously done by a coworker or contractor, while most AI output is used with little editing.

August 16, 2026
AI Is Already Taking Work Away From Coworkers Jobs

Summary: One in five employed Americans now delegates at least one task to AI that was previously done by a coworker or contractor. Software development and data analysis show the highest adoption, while 66% of AI output is used unchanged or with only minor edits. The findings point less to entire jobs disappearing than to individual tasks steadily moving from humans to machines.

workplaceaiaiautomationfutureofworkaijobsgenerativeaiaiadoption

AI Is Already Taking Work Away From Coworkers

One in five employed Americans now says AI handles at least one task that used to be done by a coworker or outside contractor.

That is the headline finding from a representative survey by Epoch AI and Ipsos, which questioned 1,106 employed U.S. adults between July 10 and 19.

The more interesting finding, however, may be what workers are doing with the output.

They're mostly just using it.

"Which parts of every job are we no longer paying humans to do?"

AI Is Moving Into Everyday Work

The survey examined 10 common workplace tasks based on U.S. Department of Labor data.

AI use was highest in software development, where 57% of workers performing that task said they use AI. Data analysis followed at 46%, while 39% reported using AI for reading work documents.

Record-keeping had the lowest adoption, at 25%.

But AI is generally not replacing an entire job.

Instead, individual tasks are moving from people to machines.

Data analysis showed the clearest example: 7.1% of respondents said AI had taken over tasks previously handled by humans. Reading work documents followed at 5.7%, with record-keeping at 5.3%.

That distinction matters.

The evidence points toward jobs being broken into smaller components, with some of those components increasingly delegated to AI.

And Workers Aren't Editing Much

Perhaps the most striking finding is what happens after AI produces something.

Epoch AI found that 66% of AI output is used unchanged or with only minor edits.

Only 6% was used completely unchanged, while 5% was heavily reworked or mostly rewritten.

That doesn't prove AI output is good. The researchers explicitly caution that editing effort isn't a measure of quality.

But it does show that AI is becoming part of the actual production process rather than simply being used as an experimental assistant.

Workers aren't necessarily asking AI for an idea and then doing the work themselves.

Increasingly, they're handing over part of the work and accepting the result.

AI Doesn't Always Save Time

More AI isn't automatically faster, either.

Workers reported saving time on 37% of tasks where AI provided partial assistance. That increased to 53% when AI handled most or all of the task.

But roughly one in six AI-assisted tasks now takes longer than it did previously.

That could mean interacting with AI introduces its own overhead. It could also mean workers are using the additional capacity to produce more detailed or higher-quality work.

The survey can't determine which explanation is responsible.

The Job Isn't Disappearing. The Task Is.

Epoch AI describes AI as a "versatile but usually not self-sufficient workplace tool."

That's probably the most useful way to read the results.

The immediate transformation isn't necessarily a world where one employee is replaced by one AI system.

It is a workplace where an employee who previously needed five people to complete a process may increasingly need four, or where those five people spend their time doing fundamentally different things.

That distinction could become increasingly important as companies redesign jobs around AI rather than simply adding AI tools to existing workflows.

What This Means for Miami

Miami's economy has a large concentration of businesses in finance, real estate, professional services, technology and other knowledge-intensive industries where individual tasks can be separated from entire jobs.

That makes this shift particularly relevant.

The question for Miami employers isn't simply "Are we replacing workers with AI?"

It may be:

"Which parts of every job are we no longer paying humans to do?"

That is a much quieter transformation, but potentially the more consequential one.

← Back to MAIN