Feature
Three years after the launch of ChatGPT, generative AI can be found in every facet of work. From drafting emails to automating workflows to analyzing data, AI tools now sit alongside employees across regions and industries.
But the rapid adoption raises a question: has work itself improved?
As organizations confront the realities of an AI-assisted workforce, the emerging picture is far more complicated than early evangelists promised. A third of US workers speak less frequently to colleagues since using generative AI and over a quarter would rather engage in small talk with an AI chatbot than a human, according to a survey of 4,000 global knowledge workers by Adaptavist Group.
Meanwhile, recent research by EdAssist by Bright Horizons of over 2000 U.S. workers found that while 59% of workers agree AI makes their job easier — up from 42% last year — 81% report pressure to take on more work and 80% feel they must deliver results faster.
The contradiction is stark. AI promises liberation but instead delivers intensification.
Has GenAI Lightened People's Workloads?
The seductive claim is that AI lightens people's workloads.
Yet the claim collapses under scrutiny. Has AI reduced workloads? "Rarely," said Bob Hutchins, founder of Human Voice Media and AI researcher. "In fact, the historical track record is that technology, almost without fail, never shortens the workday (always expands it). You 'save' a few hours of drafting emails or reports, but expectations immediately shift on output (quantity) or just lateral thinking (depth)."
Zapier saw clear gains: "Per-developer productivity rose 11% in 2025. Customer support tickets answered twice as fast," Chief People and AI Transformation Officer Brandon Sammut told Reworked. Their Customer Support's "Sidekick" cut average ticket handle time in half. Yet Sammut acknowledged a crucial distinction: "What we do see is AI reducing repetitive, low-value tasks and giving people more time for higher-impact work — not actually reducing workload,” he said.
Across sectors, the dynamic repeats. "Workload expands to the capacity to receive it. That's how innovation works," said Diane Dye, CEO at People Risk Consulting. "The rote tasks will get easier, but the complex tasks will become greater as a result."
"Expectational inflation" explains the mechanism, said Hutchins. When "stakeholders know a full report can be synthesized in minutes (as opposed to days) and the natural reaction is just to halve turnaround times," he said, the consequence is "a shrinking of the space available for human deliberation and 'slow thinking' (i.e. where creativity often lives), which is increasingly coming under pressure to match algorithmic speed."
The Bright Horizons research confirms the squeeze: 32% of workers say AI has increased pressure to learn new skills, up from 26% last year. Hours saved haven't translated to shorter days. They've been reinvested into more work, delivered faster.
Has GenAI Improved the Quality of Work Outputs?
So GenAI hasn't reduced workload. Has it at least improved quality? The evidence for this is murkier still.
The optimistic case rests on efficiency gains. According to Sammut, AI "helps teams draft solid first versions in minutes, pull insights from messy notes, and explore ideas faster than any traditional workflow." Others describe a workplace drowning in mediocrity.
"ChatGPT has made it possible for everyone to hold similar knowledge but to be really mediocre. It also gives some very bad advice," said Dye.
She stressed that while GenAI can improve first drafts, "The final draft ... still takes human work. Again, abdicate your productivity to ChatGPT and you are going to be sorely disappointed."
If productivity is used as a stand-in for quality, the answer still depends on how an organization defines productivity. "If productivity is 'volume,' then yes, for sure. But if productivity is 'value,' then we're getting to the paradox," said Hutchins. The result? "Organizations are being flooded by a sea of synthetic noise. In some sense, better productivity has become a curatorial problem: finding the needle of value in a haystack of … output."
Researchers at BetterUp and Stanford coined a term for this deluge: "workslop." Workslop is the verbose, irrelevant GenAI-generated content that has increased the workload of team members who have to sift through it.
Hutchins' team is now "working to create a workplace where we are explicit about when content is made with ChatGPT versus alone. Team members and clients get frustrated when they believe they've been handed low-effort work."
Beyond mere volume lies a deeper threat. "When we ask ChatGPT for decision support, the problem is that we're asking an echo chamber of average," Hutchins said. "The output of LLMs is not an outlier, but a calculation of the most likely continuation of an average pattern. The real risk is not hallucinations as much as homogenization, flattening the noisier, outlier thinking that is so often the seedbed of innovation."
If AI optimizes for the average, what happens to breakthrough thinking?
The Widening AI Skills Gap
The AI divide isn't just about access to technology. It's about comprehension.
The numbers from Bright Horizons are damning: 34% of employees feel unprepared for AI-driven changes, whilst 42% expect their role to change significantly due to AI within the next year.
Webinar
Dec
11
Fuel Your Front Line: Workforce Trends Shaping 2026
Get expert insight into the tools and tactics top frontline teams will need in 2026.
Register
Webinar
On demand
Fix the Disconnect Between Projects and Profit
Your margins are hiding in disconnected data. Let's find them.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
Webinar
On demand
Accelerating Healthcare Ops with AI
Sign up to see how your peers are putting AI to work — and making it stick.
Watch Now
Webinar
On demand
From Nice-to-Have to Non-Negotiable: Prove the Value of Your EX Stack
Transform your tech stack: prove its value, secure your future.
Watch Now
Webinar
On demand
The Great Workforce Breakdown (And How to Stop It)
Disconnected employees. Failing communication. It’s time to turn things around.
Watch Now
The training gap tells its own story. When employers provide AI training, adoption jumps to 76%. Without training, it languishes at 25%.
But the statistics reveal only part of the picture. What truly separates effective users from the rest isn't technical prowess alone.
Hutchins describes the distinction between the two groups as this: "Those that are approaching it as a 'magic button' with a dropdown menu of commands, versus those who are working on 'AI Literacy' as an ongoing conversation with the algorithm." True competence, he argues, "involves meta-cognition about when to not use them. The better AI we get, the more it forces organizations to realize they have to teach thinking."
Some organizations have attacked this challenge aggressively. At Zapier, Sammut said, "We've run hackweeks, embedded learning into daily work, and now assess AI fluency for 100% of candidates so new hires can contribute immediately." The results speak clearly: companywide adoption has grown from 65% in late 2023 to 97% in 2025.
But Zapier represents the exception, not the rule. "It depends on the company. Some companies do it well. And some do not. Some are purely avoidant right now," said Dye.
The consequences of avoidance are mounting. The Bright Horizons research found that 55% of workers say access to AI training or certification would make them more likely to stay with their employer. The skills gap may soon become a retention crisis — though that may prove the least disruptive of AI's effects on workplace culture.
How AI Is Effecting Human Connections
The most unsettling findings concern AI's effect on workplace relationships.
The Adaptavist research reveals tears in the workplace social fabric. A third of U.S. workers speak to colleagues less since using generative AI, whilst over a quarter would rather engage in small talk with an AI chatbot than a human. One third believe they're "addicted" to using generative AI. The research also found that while AI advances skills in content writing and clear communication, it comes "at the cost of politeness."
"AI is being used to write emails to colleagues, and those colleagues use AI to read and summarize those emails. We're running a risk of stripping out the care from collaboration: the 'Therapeutic Alliance' of the office is endangered by AI interfaces between coworkers," said Hutchins, calling this "the rise of the synthetic layer."
The problem isn't purely emotional. According to Dye, there's a competence issue at play. "What it has done is created some know-it-alls who are NOT know-it-alls. And that creates team friction."
Sammut offers a more optimistic view, framing AI as "a quiet teammate: One that preps, organizes and synthesizes." Yet even his optimism focuses on efficiency —faster collaboration with "fewer handoffs and less rework" — rather than deeper human connection.
What happens when efficiency replaces engagement?
GenAI in the Workplace: An Experiment in Progress
Three years in, GenAI in the workplace has delivered tangible gains and created problems its architects never anticipated. The effect is "net better, when used intentionally," thinks Sammut, emphasizing that "stress creeps in only when change outruns skills or guardrails."
But Hutchins' assessment cuts closer to the daily experience of many workers. "It has certainly made it more complex," he said. "[The workplace] is better only insofar as we consciously design our workflows to keep the human in the loop (we're bad at this, at the moment). If not, the world just becomes faster and noisier."
"Overall, it's a mixed bag," said Dye. "It's better in some ways, scary in others because people think they will be replaced, it's a lot of unknown. Knowing how to intelligently experiment will be the key."
As AI continues its march into every corner of the workplace, Dye's take may be the most honest assessment available: GenAI is not a revolution, but an experiment still very much in progress. The outcomes remain uncertain. The stakes continue to rise. And the question of whether work has improved remains stubbornly, maddeningly open.