Table of contents
    Was this helpful?
    Thumbs UP Thumbs Down

    Companies expected AI to boost productivity, but costly mistakes are changing their minds

    Two people are actively analyzing code and data visualizations displayed on multiple monitors.
    Table of Contents

    Companies spent heavily on artificial intelligence, expecting employees to work faster, reduce expenses, and complete more tasks. But some businesses are discovering that AI tools can also create problems. Incorrect information, repetitive documents, and rising software costs are forcing executives to reconsider how they use the technology.

    Recent reports from Reuters and The Guardian describe companies reviewing ambitious AI plans after disappointing results. Research from Ernst & Young also shows that AI failures have caused financial and operational problems for some large businesses. Rather than abandoning artificial intelligence entirely, employers are becoming more selective about where it delivers useful results.

    Companies are questioning their AI spending

    Chief software engineer conducting a quality assessment meeting with his employees.
    Source: Depositphotos

    Businesses have invested heavily in AI assistants, automated software and tools that can perform tasks with limited human guidance. Many expected these systems to reduce workloads and help employees complete projects faster. However, some organizations are finding that the financial benefits are harder to measure than expected.

    An August 2026 Fortune report described technology executives introducing limits on AI usage after spending exceeded expectations. Some companies had provided employees with widespread access to expensive tools without fully understanding how frequently they would be used or whether their output justified the cost.

    These concerns are encouraging companies to review their budgets, rather than simply buying more advanced models. Managers increasingly want evidence that AI tools save time, improve accuracy, or reduce expenses. Without those results, even widely used software may become difficult to justify.

    AI-generated work is creating additional tasks

    One concern involves employees producing documents, presentations, and reports with AI tools without checking their quality. The material may appear complete but contain errors, repeated ideas, or information that does not help colleagues make decisions. Correcting these problems can consume the time automation was supposed to save.

    Microsoft executive Ryan Roslansky described this problem in a LinkedIn post that The Guardian discussed in October. He recalled receiving an AI-generated document that offered little original thinking. After using another AI tool to summarize it, he found that the process had added no meaningful value.

    The example illustrates how productivity can be confused with the amount of material produced. An employee might quickly generate a lengthy report, but that achievement means little if coworkers must spend additional time reviewing and rewriting it. The quality of completed work matters as much as its speed.

    Meta reconsidered plans to replace workers

    Meta business office building.
    Source: Stock_Market_Visuals/Depositphotos

    Meta provides a notable example of the difficulties companies can face when introducing AI too quickly. An August 2026 Reuters investigation described an internal restructuring effort called Project OT, which aimed to reorganize parts of the company around AI agents and smaller teams.

    The plan included significant workforce reductions, with some teams facing proposals to shrink by up to 60%. However, internal problems emerged as AI tools struggled to meet productivity expectations. Employees also reported difficulties adapting to the planned changes and understanding how their responsibilities would evolve.

    Reuters reported that Meta backed away from a planned second round of broad layoffs and changed its approach to the restructuring. The company continued investing in artificial intelligence, but the experience showed that introducing automated systems does not automatically make large workforce reductions practical.

    The cost of AI tools can rise unexpectedly

    Many AI services charge businesses based on the amount of computing work their systems perform. This often involves tokens, which represent pieces of text processed by a model. More requests, longer conversations, and complicated automated tasks can increase the total amount a company pays.

    Although the price of individual AI tokens has declined, greater usage can still raise overall spending. A June 2026 Fortune report described how companies were using AI more frequently as the technology became cheaper. Running many automated tasks could therefore offset savings from lower prices.

    Some technology leaders are responding by limiting access to expensive models or directing simpler tasks toward less costly alternatives. This approach allows employees to continue using AI while reducing unnecessary spending. However, selecting cheaper software is only useful when it can complete the required work accurately.

    AI mistakes are causing serious business problems

    Server room system administrator using artificial intelligence to reduce downtime.
    Source: Depositphotos

    A September 2026 Ernst & Young survey revealed that some companies have experienced significant problems with their AI systems. The research questioned 202 senior AI executives working at organizations with at least $1 billion in annual revenue.

    According to the survey, 36% of respondents said their organizations had experienced an AI incident or failure that caused a materially negative effect. Reported consequences included financial damage, data loss, operational disruption, and reputational harm.

    The findings do not mean that 36% of all businesses experienced catastrophic failures. The research focused on a specific group of large organizations and relied on executives to report their experiences. Still, the results demonstrate why businesses need stronger safeguards before relying heavily on automated decisions.

    Companies are strengthening AI oversight

    The same Ernst & Young survey found that 98% of respondents said their organizations had formal AI governance policies. However, 47% acknowledged that their companies had previously bypassed those processes during urgent deployments. This gap between written rules and actual behavior presents additional risks.

    Among respondents whose organizations used AI agents, 26% said their companies could not detect unauthorized agents operating internally. That raises concerns about systems performing tasks without sufficient monitoring. As AI software gains access to more business tools, clear limits on its permissions become increasingly important.

    Companies can respond by conducting regular testing, improving employee training, and reviewing key AI-generated work. Human oversight remains especially valuable when mistakes could affect customers, financial records, or confidential information. These safeguards may require additional resources, but they can reduce the costs associated with preventable errors.

    Businesses are looking for measurable results

    Cropped view of businessman pointing at papers with charts.
    Source: Depositphotos

    The challenges do not mean artificial intelligence has no useful business applications. AI tools can help employees organize information, draft routine documents, and analyze large amounts of data. Their value depends on the task, the quality of their output, and the effort required to check the results.

    Business leaders are increasingly examining whether an AI system produces measurable improvements. That might include faster customer service, reduced processing time, or fewer repetitive manual tasks. Comparing those benefits with subscription fees, computing expenses, and review costs provides a clearer picture of financial value.

    For employees, the shift means using AI responsibly rather than relying on it to complete every assignment. Companies that combine suitable tools with clear expectations and human review may find more practical benefits. The current reassessment suggests that successful AI adoption requires careful planning, reliable results, and spending controls.

    TL;DR:

    • Companies are reassessing AI investments after encountering rising costs, inaccurate output, and disappointing productivity results.
    • Microsoft executive Ryan Roslansky criticized low-quality AI-generated workplace documents that added little value.
    • Reuters reported that Meta reconsidered parts of an ambitious restructuring plan involving AI agents.
    • An Ernst & Young survey found that 36% of participating executives reported material harm from AI incidents.
    • Businesses are emphasizing stronger oversight, controlled spending, and measurable productivity improvements rather than unrestricted AI adoption.

    Don’t forget to follow us for more exclusive content.

    If you liked this, you might also like:

    This article was made with AI assistance and human editing.

    Was this helpful?
    Thumbs UP Thumbs Down
    Prev Next
    Share this post

    Send feedback to ComputerUser



      We appreciate you taking the time to share your feedback about this page with us.

      Whether it's praise for something good, or ideas to improve something that isn't quite right, we're excited to hear from you.