USA 
United Kingdom
Lonely workers suffering low morale and cash being burned on wasted tokens are among the unwanted consequences of rapid artificial intelligence adoption, recent reports from leading big tech executives have revealed.
Business leaders are citing challenges with managing the behavior of their human workforce, as escalating costs and stretched budgets mount scrutiny over the billions of dollars being pumped into AI tools.
Fiona Fung, the top engineering exec on Anthropic’s Claude Code team, described how her team’s accelerated use of AI agents made work a “lonely experience” for staff.
AI fuels employee loneliness problem
Speaking on an episode of Lenny’s Podcast, Fung highlighted a problem facing employers as they rebuild and restructure teams around AI agents, in many cases stripping back the daily human interactions employees have become accustomed to.
“After a while, we felt it could start being a lonely experience because we all started just working with our agents so much,” she explained.
Fung added that the company has implemented more structured formats such as hackathons and pair programming lunches, “just to make sure we're interacting together as a team.”
She noted this is vital not only to bring staff together, but also to help them learn from one another.
Experts have warned that AI may be further fuelling the employee loneliness crisis if workers are expected to spend more time working with agents than colleagues.
‘Erode company cultures’
It is an issue particularly felt across the tech sector, where adoption is particularly rapid, with major employers axing jobs and boosting AI spending.
Last week, Oracle published an annual report showing that it shed 21,000 roles globally last year.
The “deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce,” it said.
There have been roughly 120,000 tech layoffs in 2026 so far, already closing in on the 2025 total.
A study published in Harvard Business Review found that 1,545 workers in organizations relatively far along the AI adoption curve reported feeling “lonely at work.”
The authors Constance Noonan Hadley and Sarah L. Wright warned that this workplace shift may “erode company cultures and coworker cohesion in the future.”
According to Gallup, around one in five employees globally report feeling lonely at work, while up to 58% say they experience loneliness at least some of the time. Younger workers aged 18-24 are particularly vulnerable and are twice as likely to feel lonely as older colleagues.
Meta is another business to reduce the size of its workforce in recent months – by an estimated 8,000 roles – while citing greater AI spending.
Turning workforce data into early warnings for high-cost employees
The company has since struggled with an employee morale crisis. Andrew Bosworth, Meta’s Chief Technology Officer, admitted in an internal memo that “entire teams” were “left in the lurch,” while staff complained that changes left them with minimal interaction with human colleagues.
The tokenmaxxing risk
A further issue facing employers is how bulging AI budgets are being spent by staff.
Tokenmaxxing – a trend in which workers are burning through AI tokens to signal productivity and volume of usage – shows how pressure to use tools may not translate into desired results.
At Accenture, for example, some non-tech workers are reportedly spending tokens on tasks such as converting PDFs into PowerPoints.
“We’re seeing from some of the data internally at least that it’s actually not our engineers that are driving the token consumption,” Accenture’s Head of AI Strategy Justice Kwak said in an internal meeting, the transcript of which was obtained and reported by 404.
Another firm, Slash, also claimed to be dealing with the problem – albeit in an apparent social media campaign. One worker reportedly spent $80,000 in tokens to vibecode a “brainrot” game.
The problem is very real for many employers, however, with widespread challenges in measuring return on investment for mounting AI costs.
