A groundbreaking new analysis of a comprehensive survey, published by the National Bureau of Economic Research and highlighted by Fortune, reveals a stark reality: despite widespread enthusiasm and significant investment, Artificial Intelligence has, for the vast majority of businesses, yet to deliver on its promise of boosting productivity or transforming employment. This extensive survey, encompassing nearly 6,000 top executives – including CEOs, chief financial officers, and other C-suite leaders – across major economies like the US, UK, Germany, and Australia, paints a picture far removed from the utopian visions often espoused by tech evangelists. The overwhelming consensus, voiced by approximately 90 percent of those interviewed, is that AI has made no discernible impact on either productivity or employment within their organizations.

This finding is particularly striking given the high rate of AI adoption. The survey indicates that around 70 percent of these firms are actively integrating and utilizing AI technologies in their operations. This means that a significant majority of companies that have taken the plunge into AI are openly admitting that the technology has, so far, failed to move the needle. It raises profound questions about the current state of AI implementation, the maturity of the technology for business applications, and the often-inflated expectations surrounding its capabilities.

The executives themselves, despite being the driving force behind AI adoption, appear to derive surprisingly little personal benefit from these advanced tools. While two-thirds of the surveyed leaders reported personally engaging with AI, their average usage amounted to a mere 1.5 hours per week. This is less time than many individuals spend passively scrolling through social media on their phones in a single day, suggesting that even at the highest echelons of corporate leadership, AI is not yet deeply integrated into daily workflows or decision-making processes. This limited personal engagement is even more perplexing when juxtaposed with executives’ generally more optimistic outlook on AI compared to their subordinates. Another recent survey, for instance, illuminated a significant "perception gap": while 98 percent of bosses believed AI saved their employees time, a substantial 40 percent of rank-and-file white-collar workers disagreed, feeling it offered no time-saving benefits whatsoever. This disconnect suggests that the strategic vision for AI may not be translating effectively into tangible, on-the-ground improvements for those directly using the tools.

These latest findings add considerable weight to a growing body of evidence questioning AI’s actual economic impact and its purported ability to supercharge workplace productivity. The notion of an "AI productivity paradox" is gaining traction, echoing the "Solow paradox" of decades past, where Nobel Prize-winning economist Robert Solow famously remarked that "you can see the computer age everywhere but in the productivity statistics." Like the early days of information technology, AI’s transformative potential has been widely touted, yet concrete, measurable economic gains remain elusive. Prior surveys have also underscored this skepticism; more than half of nearly 4,500 CEOs in another recent study admitted their companies weren’t seeing a financial return on their AI investments. Perhaps even more alarming was an MIT study that rang alarm bells across the industry, reporting that a staggering 95 percent of companies that integrated AI experienced no meaningful growth in revenue. These figures collectively paint a sobering picture for a technology that has attracted billions in investment and is frequently presented as the next industrial revolution.

The reasons behind this underwhelming performance are becoming clearer as research delves deeper into the practical application of AI in the workplace. Far from being a universal panacea, studies have consistently found that current AI models frequently fall short when tasked with complex remote work or nuanced white-collar responsibilities. Their limitations become apparent in situations requiring critical thinking, creative problem-solving, or a deep understanding of human context and emotion. For instance, in the realm of software development, a field often highlighted as ripe for AI disruption, the technology has been found to slow down rather than speed up human programmers. This counterintuitive outcome often stems from AI’s propensity to generate code with subtle errors or inefficiencies, necessitating significant human oversight, debugging, and refactoring – ultimately adding to the workload rather than reducing it.

Beyond mere inefficiency, a fresher and arguably more damning avenue of research is exploring AI’s profound and often negative effects on the workforce itself. One report revealed that the technology may be inadvertently intensifying work and accelerating employee burnout. The expectation that AI can handle routine tasks can lead to a paradoxical increase in human workload, as employees are then tasked with higher-level, more complex problems, or are required to constantly monitor and correct AI outputs. This creates a cycle of increased pressure and cognitive load. Furthermore, another study coined the term "workslop" to describe the low-quality, error-ridden output frequently produced by AI, which then falls to human co-workers to fix. This "workslop" not only bogs down workflows and wastes valuable time but also breeds resentment among team members who are forced to clean up after an imperfect AI. The researchers behind this insight ominously concluded that AI’s "most alarming cost may be interpersonal," highlighting the erosion of trust, collaboration, and morale within teams grappling with these new digital assistants.

Despite these significant hurdles and widespread reports of negligible impact, the adoption of AI continues its upward trajectory. The new survey notes that the percentage of businesses using AI technology climbed from 61 percent in the February-April 2025 period to 71 percent between November 2025 and January 2026. This paradoxical trend suggests that a combination of factors – perhaps a fear of being left behind, the allure of future potential, or sustained pressure from investors and market competitors – is driving companies to embrace AI, even in the absence of immediate, quantifiable returns. It underscores the immense hype cycle surrounding AI, where the promise often outweighs current reality, and long-term vision frequently overshadows short-term disillusionment.

The historical parallel to the Solow paradox remains a potent lens through which to view AI’s current state. When information technology first emerged, it similarly failed to translate into an immediate surge in productivity, leading to a period of slow growth despite obvious technological advancements. It took decades of societal and organizational restructuring – re-engineering business processes, developing new skills, and adapting corporate cultures – before the true economic benefits of computers were widely realized. The question for AI, then, is whether it is merely in an analogous early phase, awaiting similar systemic adjustments, or if its fundamental limitations, particularly in areas requiring human-like judgment and creativity, will prevent it from ever fully delivering on the expansive promises made by its proponents.

Interestingly, despite the current lack of impact, the surveyed executives are not abandoning hope. They project that AI will eventually boost productivity by 1.4 percent and overall output by 0.8 percent over the next three years. However, this optimism comes with a notable caveat: they also anticipate a corresponding 0.5 percent reduction in employment. This projection reveals a complex calculus within boardrooms, where the anticipated benefits of AI are intrinsically linked to potential job displacement. It raises crucial ethical and societal questions about the true cost of these predicted gains and which "part they’re excited about more" – the modest productivity increase or the significant reduction in labor costs. The journey of AI in the workplace is clearly far from over, but the initial data suggests a bumpy road ahead, laden with challenges that extend far beyond mere technological integration.

More on AI: Oxford Researcher Warns That AI Is Heading for a Hindenburg-Style Disaster