Companies Replaced Workers With AI; Now 75 Percent Say It Cost Them More

The promise of artificial intelligence was simple. Replace some human workers, lower payroll costs and make businesses more efficient. For many companies, the calculation has proved less straightforward.

According to new industry data reported by ZDNET, 75% of organisations that replaced employees with AI found that the move cost more than it saved. The finding highlights a growing challenge for executives under pressure to show a return on their AI investments.

The issue is not that AI cannot perform useful work. It is that replacing people can create costs that are easy to overlook when an automation project begins.

The Hidden Price of Replacing Workers

An employee's salary is only one part of the cost of running a business. AI systems require software, computing capacity, security, and ongoing maintenance. They also need people to monitor performance, correct errors, and update the systems as business needs change. Those requirements can reduce the savings expected from job cuts.

Companies can also lose institutional knowledge when experienced employees leave. Workers often understand customers, internal systems, and unusual problems that are difficult to capture in software. That knowledge can become particularly valuable when an automated system encounters a situation outside its normal pattern.

When Automation Creates More Work

The problem becomes clearer when AI is deployed in roles that involve complex decisions or customer interaction. A chatbot may answer routine questions quickly. It may struggle with a complaint that requires context or discretion.

An automated financial system can process large volumes of applications. It may still require experienced staff to examine unusual cases.

The same principle applies across many industries. Automation can handle predictable tasks well, while complicated situations often continue to require human judgement. If companies remove too many workers, they can find themselves paying later to restore the expertise they cut.

The Wider Data Is More Complicated

The 75% finding does not mean AI is reducing employment across the entire economy. Research from the National Bureau of Economic Research offers a broader perspective. A survey of almost 6,000 senior executives across the US, UK, Germany, and Australia found that more than 90% of businesses reported no effect from AI on employment during the previous three years.

About 89% also reported no effect on labour productivity during that period. Executives nevertheless expect AI to deliver productivity gains in the years ahead. That suggests businesses are still experimenting with the technology. Many may be investing today in systems they expect to produce larger benefits later.

The Shift from Replacement to Support

That uncertainty is encouraging some companies to rethink how they use AI. Instead of asking how many jobs a system can eliminate, businesses can ask how much more work employees can complete with AI assistance. The distinction is important.

AI can search documents, summarise information, identify patterns, and deal with repetitive tasks. Employees can then spend more time on decisions that require experience and judgement.

Research by Express Employment Professionals and Harris Poll found that 90% of US hiring managers surveyed believed AI would not eliminate the need for employees at their companies. The survey also found that 92% said their companies remained committed to preserving a human element in the workplace. The findings suggest that AI and employment are not necessarily opposing forces.

A Different Way to Measure Success

The latest figures may force executives to reconsider how they calculate AI's value. Headcount reductions are easy to measure. Productivity is more complicated.

A company might eliminate a number of positions but then spend heavily on AI infrastructure, specialist staff, and additional oversight. It could also face higher costs if service quality falls or employees must later be rehired.

A different model is to keep skilled workers while using AI to increase their output. That approach may not produce an immediate reduction in payroll. But it can allow businesses to handle more work without sacrificing human judgement.

The Lesson from the First AI Wave

The 75% figure should not be treated as proof that AI has failed. It is a warning against assuming that automation automatically produces savings. Companies still have strong reasons to adopt AI. The technology can process information quickly and take over repetitive tasks. But the financial benefits depend on how it is introduced and managed.

For businesses that rushed to replace workers, the lesson may be particularly costly. AI can reduce the need for some tasks. It does not necessarily remove the need for people who understand the business, its customers, and its risks.

The next phase of AI adoption may therefore be less about replacing workers and more about making them more productive. For executives, that could mean a different question is now worth asking: not how many people can AI replace, but how much more can those people achieve with it?