Worker frustration with AI is rising as management optimism holds

AI enthusiasm in the workplace is splitting along class, role, and demographic lines

As artificial intelligence becomes more embedded in everyday work, employee attitudes appear to be moving in the opposite direction from executive rhetoric. A Glassdoor-based analysis cited by The Decoder suggests that positive sentiment about AI among U.S. workers has fallen sharply since 2019, while negative commentary has overtaken it.

The shift is notable not simply because AI remains one of the dominant business narratives of 2026, but because the data points to widening differences in how workers experience the technology. Leaders and some technical roles still describe AI positively. Many frontline and administrative workers do not. The result is a labor sentiment split that says as much about implementation and power as it does about the tools themselves.

According to the source text, positive AI sentiment in worker reviews dropped from 81% in 2019 to 43% by mid-2026. Over the same period, negative comments rose to 53%. Mentions of AI also surged, with reviews referencing terms such as “AI,” “LLM,” “GPT,” “Artificial Intelligence,” or “OpenAI” increasing from under 0.2% of reviews in 2019 to more than 2.3% by May 2026. From May 2025 to May 2026 alone, AI mentions jumped 240%.

More discussion, less goodwill

That combination matters. Employees are not simply talking about AI more often because it is fashionable. They are talking about it more because it is showing up across professions, management processes, and daily workflows. And as exposure increases, the tone of those discussions appears to be worsening.

The source text suggests that jobs with very little AI exposure, such as butchers or electricians, rarely mention the topic. But once occupations reach even a moderate level of AI exposure, comments begin appearing across a wide range of roles, regardless of whether the work is seen as highly automatable. That implies a broad workplace effect: AI is becoming a management and process issue, not just a software feature.

The analysis also indicates that positive experiences tend to cluster in just two patterns. First are employees at companies that benefit directly from the AI boom, described in the source text as “AI winners.” Second are workers who report that AI removes repetitive tasks and gives them more autonomy or efficiency. Outside those conditions, dissatisfaction appears easier to generate than enthusiasm.

Managers and executives see a different AI than staff do

The strongest divide in the findings is organizational. Leadership roles are described as the most positive about AI, especially in reviews from June 2025 through May 2026. Executives are identified as the most optimistic group overall. By contrast, several worker categories register overwhelmingly negative sentiment.

Insurance claims workers rank among the harshest critics, with the source text saying their comments are almost entirely negative. These employees commonly complain that management is forcing buggy tools into workflows. Writers, accountants, and customer service representatives also skew strongly critical, in some cases by margins of roughly four to one.

Even within the technology workforce, the split is not uniform. Software architects mention AI more than any other role and still tend to evaluate it positively. Software engineers, whose work more directly involves writing and reviewing code, are said to be much more skeptical, with 57% of their AI-related comments trending negative.

That distinction is revealing. Workers closest to implementation may have a clearer sense of the mismatch between product promise and operational reality. Meanwhile, more senior decision-makers may be evaluating AI at the level of strategy, investor expectations, or productivity potential rather than day-to-day friction.

The demographic warning sign

The source text also points to sharp demographic differences, especially among younger women. It says only 21% of AI-related comments from Gen Z women are positive, making them the most skeptical group in the analysis. That is a striking signal because it suggests disillusionment is not confined to older workers worried about technological disruption. Some of the strongest skepticism is showing up among workers who entered the labor market during the generative AI buildout.

While the supplied text does not provide a full causal explanation, it does make one broader implication difficult to ignore: the experience of AI at work is not being distributed evenly. Role, status, and demographic position appear to shape whether AI is encountered as a useful assistant, a source of pressure, or a poorly functioning system imposed from above.

For companies, that should shift the conversation. The issue is no longer whether employees have heard of AI or whether leadership teams want to deploy it. The issue is whether organizations can implement it in ways that workers regard as competent, fair, and genuinely helpful.

A management problem as much as a technology story

The most important lesson in the analysis may be that employee backlash is not reducible to simple automation fear. The source text explicitly says complaints go beyond job-loss concerns. Workers appear frustrated by buggy tools, top-down mandates, and implementations that add friction instead of removing it.

That puts pressure on one of the most common narratives in enterprise AI adoption: that resistance mainly reflects unfamiliarity or anxiety about change. If sentiment keeps deteriorating as use spreads, companies may need to confront a less convenient explanation. Some deployments may be generating real workplace costs that are visible to staff long before they show up in executive dashboards.

There is still room in the data for a more positive trajectory. Employees do respond well when AI reduces repetitive work or when they are positioned inside firms benefiting materially from the boom. But those paths to satisfaction appear narrow compared with the many routes to disappointment.

For a technology that has been sold as a universal productivity layer, that asymmetry matters. The next phase of enterprise AI may depend less on bigger models and more on whether companies can close the credibility gap between what management thinks AI is doing and what workers say it is actually doing.

• Positive AI sentiment in U.S. worker reviews fell from 81% in 2019 to 43% by mid-2026.
• Negative AI comments rose to 53%, while AI mentions in reviews jumped sharply.
• Executives and managers remain the most optimistic groups.
• Insurance claims workers, writers, accountants, and customer service staff are among the most critical.

This article is based on reporting by The Decoder. Read the original article.