Post by : Rameen Ariff

A new study from the Massachusetts Institute of Technology (MIT) has revealed a surprising truth — most companies are not gaining any real benefits from their big spending on artificial intelligence.

For the past few years, firms across the world have been rushing to bring in smart computer systems with the hope of cutting costs and increasing profits. But according to MIT’s latest report, 95 out of every 100 projects have shown no return on investment.

The report, titled “The GenAI Divide: State of AI in Business 2025”, found that despite the billions being poured into such projects, only a very small number are proving successful.

Billions Spent, Almost Nothing Gained

Researchers said that between $30 to $40 billion has been spent by large companies to bring in advanced systems. Yet, the study showed that only five percent of these projects are making real money, while the rest are stuck without any clear impact on profits or loss.

The research looked at 300 company projects and included talks with around 350 workers. Tools such as ChatGPT and Copilot are among the most widely used, but experts noted that most of the time these tools are only helping individual workers finish tasks faster. They are not creating big changes in the company’s financial numbers.

One part of the report said, “Over 80 per cent of organisations have explored or piloted these tools, and nearly 40 per cent report deployment. But these tools primarily enhance individual productivity, not profit and loss performance.”

Why Are Projects Failing?

The MIT team explained that the main reason for this failure is not the weakness of the technology itself. In fact, many of these systems are powerful. But the problem lies in how hard it is to fit them into the daily work style and systems of companies.

Most businesses already have long-established ways of working, and it is proving difficult to mix the new systems with these older methods. There is also a “learning gap” — many employees are struggling to understand how to use the new tools properly. While company executives are often quick to blame the performance of the technology, researchers suggest the real issue is poor adoption and training.

Example: Taco Bell’s AI Drive-Through

A recent example comes from Taco Bell. The company had been testing voice-based systems in its drive-through outlets. But the Chief Digital and Technology Officer, Dane Mathews, admitted that they had to slow down the rollout.

He said the technology was not always helpful and, in some cases, slowed down service. According to him, during very busy hours, human workers are still better at taking orders quickly and correctly.

Mr. Mathews added, “For our teams, we will guide them: at your restaurant, at these times, we recommend you use voice systems, or at other times, we suggest you monitor them closely and step in when needed.”

This reflects a wider reality: while these new systems can be impressive, they still cannot always match the flexibility and quick thinking of human staff.

Apple’s Warning on Smart Systems

Adding to the debate, tech giant Apple published its own study in June, which cast doubt on the actual “intelligence” of many modern systems.

The report, called “The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity”, argued that these systems are not truly reasoning. Instead, they are very good at memorising and spotting patterns.

Apple explained that when a question is simple and follows known patterns, the systems perform well. But when the problem becomes more complex, or when the question is changed in an unexpected way, the systems “collapse”. In other words, they may look clever on the surface, but they cannot think like a human.

The Bigger Picture

The findings from MIT and Apple raise tough questions for businesses. On one hand, companies are eager to show they are modern and forward-looking by investing in new technology. On the other, most of these investments are not leading to real value — at least not yet.

For many firms, the rush to adopt has led to unfinished experiments, with tools being tested but not fully fitted into work systems. Experts say that unless companies invest not only in the systems but also in training their workforce and reshaping workflows, the money spent will keep going to waste.

The MIT report suggests that instead of chasing big headlines, companies may need to move slower, focus on small wins, and ensure that new tools are serving their actual needs rather than just being used as a trend.