When Every Word Becomes Data: AI Rewrites the Workplace

Every phone call can now become a scorecard. At the Co-op, AI is listening to conversations between legal services staff and customers seeking help with probate, wills and estates, opening a new fight over workplace trust, stress and control.

The change reaches beyond one employer. AI is moving into monitoring, reporting and management itself, changing what workers are expected to do and forcing organizations to ask a bigger question: when systems track every action and coordinate digital labor, what is left for managers and employees to own?

The Co-op Turns Calls Into Performance Scores

Co-op Legal Services has introduced automated listening technology to dozens of legal services staff in the last couple of months. The system records and analyses calls, then awards a percentage score for interactions between agents and customers who need advice about probate, wills and estates.

An OpenAI model assesses more than 50 discrete aspects of each phone call. It provides pass and fail scores to managers, who use the data to analyse employee performance. For one worker, that can mean AI monitoring their work for seven hours a day while they handle inquiries from customers who are often recently bereaved.

A whistleblower described the system as “oppressive and dystopian”. One Co-op worker said, “You are being watched,” and explained that the technology creates distrust and a sense that the workplace is dystopian.

“It’s different from occasionally being listened to for compliance purposes. [You are] being monitored absolutely every word that you say and every time that you utter something … It just feels really dystopian. It’s the thin end of the wedge,” the worker said.

The criticism focuses on more than software. Workers are questioning whether an automated score can capture the demands of sensitive conversations, especially when customers are dealing with bereavement and complex legal issues.

The Co-op said it did not recognize the criticism and that its probate advisers were highly engaged. It said AI supports colleagues as a key part of quality assurance and helps ensure clients receive empathetic guidance.

Experts have voiced fears that expanding AI monitoring will increase workplace stress. John Chadfield said, “AI is sold as the solution to every problem, but examples like this make it clear that it can diminish sensible workplace practices and intensify working lives for no reason.”

Monitoring Is Spreading Across Workplaces

The Co-op is the latest employer in a wider shift toward AI surveillance and automated oversight. Euan Blair’s AI training company, Multiverse, has started using AI for blanket monitoring of online classroom sessions, with teachers calling the system “remorseless” and “unnerving”. Multiverse said it was “using AI to improve the experience of learners and customers”.

Other experiments have already exposed the tension between measurement and trust. In February, Burger King announced it would use AI to track restaurant workers’ customer interactions. In summer, Meta paused tracking employees’ computer keystrokes.

Worker pressure is also showing up in how people describe their own AI skills. In a Visier survey, 45% of workers felt pressure to use AI even when they were not confident about doing so effectively, while almost half of US employees said they had exaggerated their AI use or expertise to colleagues or leadership.

That pressure creates a workplace feedback loop: companies add AI to measure performance, workers feel watched or pushed to demonstrate AI expertise, and managers receive more data to interpret. Trade unions and experts are raising concerns about workplace stress as these systems spread.

AI Is Rebuilding the Manager’s Job

The other side of this transformation is not surveillance but management redesign. Wendy Smith reported that the manager role is being reassembled around a different mix of responsibilities, with some management work moving into systems while other duties expand.

At Edward Jones, CIO Kevin Adams created an AI system that tracks context across enterprise programs and committees. The system builds context within each environment by adding relevant documents, decisions, history, open issues and commitments over time.

At CVS, agents now prepare reports and dashboards that once formed part of management meetings. That change reduced meetings from two hours to roughly 15 minutes, showing how AI can remove layers of work built around collecting, repackaging and transmitting information.

AI responsibilities can include monitoring, reporting, preparation and recurring operating rhythms, reducing active managerial labor. Removing administrative work can give managers more capacity for coaching, mentoring and developing organizational capabilities.

But saving time does not automatically change a role. If AI saves a manager five hours and an organization fills that time with more projects or meetings, capacity improves without changing the work’s basic structure. The changes at CVS and Edward Jones therefore raise a tougher question: which management layers exist because organizations need leadership, and which exist because information has been difficult to share?

Some management responsibilities are expanding instead. Managers may need to assign tasks to digital labor, structure workflows, inspect outputs, coordinate agents, resolve failures and train agents. Alan Rosa at CVS described his new balance in clear terms: “I spend 70% of my time training it and 30% of my time using it.”

The old line between managers and individual contributors is also blurring. AI lets people without direct reports perform management activities for digital labor, while the responsibilities of developing people, overseeing agents and owning workflows remain related but not interchangeable.

Tyler Derr at Broadridge is considering a future in which some frontline managers may manage almost exclusively agents. Broadridge is thinking about how to identify agents, assign work, monitor behavior and determine controls across the company. A manager responsible for three employees and 19 agents would work across different spans: human span, agent span, workflow span and total productive leverage.

IBM’s CIO for Technology Platform Transformation, Matt Lyteson, discussed management shifting toward attaching more directly to outcomes. That shift could make information-focused layers harder to justify, while increasing the value of people who can guide systems, develop workers and take responsibility when automated processes fail.

The workplace is moving into a contested phase of AI adoption. Systems can shorten meetings, organize knowledge and support quality assurance, but they can also turn every word into a performance signal. The next challenge will be deciding whether AI gives workers and managers more room to do meaningful work—or simply creates new ways to watch, score and pressure them.

Stay connected via Google News

Follow us for the latest AI News