AI Overuse Is Quietly Eroding the Thinking Skills That Got Workers Hired in the First Place

A new study finding that ten minutes of AI use degrades independent thinking lands against a backdrop of Indian tech workers already reporting burnout and H-1B anxiety — raising questions about whether the productivity gains from AI tools are being paid for in cognitive currency.

By Moemedi Michael Poncanasouth-asia4-minute read7 May 2026☆ Save ↗ Share ⎙ Print

The evidence keeps building in the wrong direction. A peer-reviewed study published in early 2026 found that just ten minutes of AI-assisted work measurably degrades a person's ability to solve problems independently — a finding that lands with particular force in a labor market where Indian tech professionals on H-1B visas are both the heaviest adopters of AI tools and the most exposed to the consequences of cognitive erosion. Workers on these visas operate under compounding pressure: employers demand high-output productivity, clients expect near-round-the-clock availability across time zones, and the constant churn of project work rewards those who delegate thinking to the AI du jour. The very conditions that make AI adoption attractive are, the study suggests, the conditions most likely to hollow out the skillsets that made those workers employable to begin with.

The H-1B workforce in the United States is substantial — Indian nationals have comprised the largest share of visa recipients in recent years — and it occupies a specific position in the tech labor hierarchy. Workers in this cohort are not merely competing in a market; they are competing under the constraint that their legal status is tied to employer sponsorship. A layoff is not just a professional setback. It is, as one Indian professional working with US clients told The Indian Express, a countdown. "I just count the days left in the US," he said, describing the psychological state of workers who know their employability is their immigration status. That precarity does not naturally incline toward measured, critical engagement with the tools one uses. It inclines toward speed, output, and doing what the system rewards.

The study's findings complicate the prevailing industry narrative. Tech companies have positioned AI tools as productivity multipliers — copilots that handle the routine so the human can engage the complex. That framing is not wrong, exactly. Tasks get done faster. Inboxes get cleared. Drafts get written. But the study suggests that the cognitive complexity that remains — the part the worker is supposed to handle — gets processed differently after a period of AI delegation. The independent problem-solving muscle appears to atrophy under conditions of heavy tool-reliance, even when the tool use is intentional and task-appropriate. Whether this effect is temporary or cumulative, reversible or permanent, the study does not yet establish. What it does establish is that the question deserves serious attention, and that the populations most likely to be testing its answers are the ones with least institutional power to demand that the question be asked.

The burnout data reinforces the frame. The same Indian Express reporting that documented H-1B anxiety also chronicled a pattern of overwork among engineers working with American clients — a cohort including NIT alumni reporting that sleeping at 4am left them physically and cognitively depleted. The connection to AI is not explicit in those accounts, but the structural logic is not hard to trace. When a worker is already operating at the limits of sustainable effort, reaching for an AI tool is not a considered choice. It is a reflex. And reflexes, it turns out, may be trading short-term output for long-term capability.

The counterargument is not without merit. AI tools demonstrably reduce error rates in routine coding tasks, accelerate documentation workflows, and allow smaller teams to produce outputs that previously required larger headcount. For workers on tight timelines with high output expectations, those gains are real. The question is not whether AI tools deliver value — they do — but whether the current deployment paradigm, which rewards speed and delegation without accounting for cognitive side effects, is sustainable for the workers doing the delegating. The industry has a strong interest in not asking that question. A workforce that is aware of the cognitive tradeoffs being made on its behalf would likely demand compensation, flexibility, or structural change. The productivity narrative serves employers' interests better than the cognitive-erosion narrative does.

What makes this structurally significant is the compounding effect. Workers who use AI most heavily are, the study suggests, the ones most likely to lose the independent reasoning capacity that made them valuable in the first place. If that capacity is partially degrades over time, the workers most dependent on AI tools become simultaneously more dependent and less replaceable in the ways that matter — more replaceable in the task-execution sense, less replaceable in the problem-solving sense. The effect, if it holds, inverts the productivity story. Instead of AI augmenting human cognition, it progressively displaces the human capacity it appeared to complement. The workers least able to question their tool-use patterns — those facing visa pressure, performance metrics, and round-the-clock client demands — are the ones most exposed to the downside.

The study does not answer whether this is fixable. It does not test whether deliberate,间歇性的 disengagement from AI tools — structured periods of unaugmented problem-solving — can reverse or prevent the effect. Those are open questions. What the research establishes is that the assumption embedded in the industry's productivity narrative — that AI use is costless to the human cognitive substrate — is false. The cost is real, and it is being paid most heavily by the workers least positioned to negotiate its terms. Whether the industry that benefits from their output will eventually care about the cognitive price its workers are paying is a different question. The evidence now says that question deserves an answer.

The thread context for this piece drew on Indian Express reporting on H-1B precarity, tech worker burnout, and AI cognitive research published in early 2026. Monexus notes that while Western wire coverage of AI in the workplace tends to emphasize productivity gains for employers, the worker-experience angle — including cognitive side effects and labor precarity — receives notably less column-inches in the general business press.