Executives once sold generative AI as the fix for overstretched technology groups. It would automate the drudgery. It would free engineers to tackle strategic work. Productivity would surge. Yet many IT departments now report the opposite. They face more alerts, more tools, more outputs to verify. The mental tax has grown heavier.
TechRadar recently captured this shift in sharp detail. Its analysis draws on fresh research showing that three-quarters of IT professionals say AI has made their roles more demanding. Only one in five see any real drop in mental effort. The piece, written by SolarWinds chief product officer Cullen Childress, labels the core problem “signal overload without context.” AI systems pour forth insights, flags, and suggestions. Few arrive with the background needed to act on them quickly. The result feels like noise, not clarity. TechRadar
Childress calls the extra validation step a “trust tax.” Hallucinations persist even in strong models. IT staff sit on the front lines of security, compliance, and system stability. They cannot afford to trust blindly. So every recommendation carries an implicit cost in time and judgment. That cost accumulates fast across multiple platforms.
Tool sprawl makes matters worse. Organizations adopt several AI assistants with overlapping features, different interfaces, and inconsistent results. Engineers switch contexts constantly. Each switch drains focus. The very technology meant to simplify ends up fragmenting attention further.
But the issue runs deeper than interface fatigue. New studies document how AI changes the nature of work itself. A February 2026 Harvard Business Review article argues that AI does not reduce workloads. It intensifies them. Employees work faster, take on broader responsibilities, and stretch hours without being asked. The initial productivity burst often gives way to fatigue, lower-quality output, and higher turnover risk. Harvard Business Review
Researchers at Boston Consulting Group identified a related condition they term “AI brain fry.” In a survey of nearly 1,500 full-time workers, those managing four or more AI tools saw productivity collapse. Thirty-four percent of those reporting the symptoms said they intend to quit. The phenomenon brings 14 percent more mental effort, 12 percent more fatigue, and 19 percent more information overload. A Substack analysis by Dr. Philippa Hardman ties these figures to broader patterns, including the rise of “workslop” — polished but shallow AI content that demands hours of human rework. Dr. Philippa Hardman’s Substack
Psychologists add another layer. Removing routine tasks may strip away exactly the mental breaks brains need to recover. Amy Morin, psychotherapist and author of “The Mental Strength Playbook,” warns that constant high-level focus depletes bandwidth. “We only have so much attention and so much mental bandwidth,” she told Fortune. “And if we’re doing these high-level tasks all day long, we’re going to run out of energy way faster.” A University of Texas at Austin study found that brief low-effort pauses every five minutes boosted productivity by more than 7 percent. AI that eliminates those pauses may accelerate burnout. Fortune
Anthropic’s research, cited in the TechRadar piece, shows the double edge clearly. AI can speed certain coding tasks by up to 80 percent. Over time, however, it risks eroding the deep problem-solving skills that experienced engineers rely on. The long-term effect looks less like augmentation and more like atrophy in key cognitive muscles.
Upwork’s earlier data revealed a telling gap. Ninety-six percent of executives expected productivity gains from AI. Seventy-seven percent of employees said it actually increased their workload. Thirty-nine percent now spend more time reviewing AI-generated material. Seventy-one percent report burnout. These numbers have not improved in subsequent surveys.
So what separates teams that gain from those that drown? The successful ones consolidate rather than expand their tool sets. They pick a few well-integrated platforms and stick with them. They set explicit governance rules that spell out when AI can act alone, when humans must review, and how outputs get validated. They treat training as a prerequisite, not an afterthought. And they prioritize observability across hybrid environments before layering intelligence on top.
Childress puts it plainly. “The most effective teams are consolidating, not expanding.” They become orchestrators instead of mere operators. They build guardrails first, then scale within them. Without that discipline, AI simply accelerates existing chaos.
Recent commentary echoes the warning. A California Management Review article from early 2026 describes the productivity blind spot. AI shifts cognitive constraints rather than removing them. The mental work moves from creation to constant verification and integration. Technostress rises. Decision quality can suffer.
Microsoft researchers reached similar conclusions in their study of knowledge workers using tools such as ChatGPT and Copilot. Greater confidence in the AI correlated with less critical thinking. The effort shifts from gathering information to verifying it, from solving problems to stitching AI responses into coherent work. Over-reliance risks long-term skill loss.
None of this suggests AI lacks value. Used with care, it handles repetitive analysis, surfaces patterns, and accelerates routine code. The danger appears when adoption outruns thoughtful integration. When leaders chase the next shiny model without addressing data quality, governance, or human limits.
IT teams sit at the sharp end of this experiment. They manage the complex, distributed systems that power modern business. They face pressure to deliver faster while maintaining security and reliability. Adding unstructured AI output into that mix does not lighten the load. It multiplies the variables they must juggle.
The data paints a consistent picture. Initial time savings get eaten by rework, validation, and context switching. Mental fatigue climbs. Some workers thrive in the new environment. Many others report feeling overwhelmed. A small group — around 10 percent in one breakdown — achieves sharper thinking and higher engagement. The rest navigate varying degrees of strain.
Organizations that ignore these signals risk higher attrition, especially among skilled technical staff. They risk poorer system decisions made under cognitive pressure. And they risk squandering the very productivity gains they hoped to capture.
The path forward requires restraint. Pick tools deliberately. Govern their use rigorously. Train people to direct AI rather than chase its suggestions. Measure outcomes by net human effort, not gross output. Simplify the underlying environment first. Only then does intelligence deliver on its earlier vows.
AI will keep evolving. Its outputs will grow more convincing. The trust tax may shrink with better models. Yet human cognitive capacity stays finite. The organizations that respect that limit while still extracting value stand to pull ahead. Those that treat mental bandwidth as infinite may find themselves managing exhaustion instead of innovation.
WebProNews is an iEntry Publication