The stronger AI becomes, the more exhausted people feel—“anxiety” has become the norm for both companies and employees.
TechFlow Selected TechFlow Selected
The stronger AI becomes, the more exhausted people feel—“anxiety” has become the norm for both companies and employees.
AI was supposed to be a labor-saving tool, yet it has become a new source of stress in many workplaces.
Navigating Web3 tides with focused insightsBy Xu Chao
Source: WallStreetCN
AI-powered programming tools promise to liberate engineers—but in reality, they’re fueling a new wave of productivity anxiety.
As AI coding agents like Anthropic’s Claude Code and OpenAI’s Codex continue to advance rapidly, tech companies are falling into an organization-wide “productivity obsession”—driven from the top down. Executives are personally writing code; employees are pressured to increase their frequency of interaction with AI; and overtime hours aren’t shrinking—they’re growing. Though AI was meant to be a labor-saving tool, in many workplaces it has become a new source of stress.
Survey data reveals a stark cognitive gap: consulting firm Section found that over 40% of C-suite executives believe AI tools save them at least eight hours per week, while 67% of non-managerial employees say AI saves them less than two hours—or none at all. A longitudinal study by UC Berkeley involving a 200-person organization further found that even as employees offload large volumes of work to AI, their actual working hours continue to lengthen.
This anxiety is rooted in structural forces. When CTOs are coding with AI at 5 a.m. and CEOs measure team effort by invoice amounts, the industry’s very definition of “efficiency” has been rewritten—and the cost of that redefinition is being borne by ordinary employees.
Executives Enter the Coding Arena—Productivity Anxiety Spreads Top-Down
The term “vibe coding” originally carried a laid-back promise. Introduced to the public in February 2025 by former OpenAI researcher Andrej Karpathy, it described a new programming paradigm where engineers simply chat with AI to complete development—“fully immersed in the vibe.”
Yet one year later, the vibe has shifted entirely.
Alex Balazs, CTO of Intuit, describes his recent routine: His wife comes downstairs at 8 a.m. and finds him already deep into work for several hours. “She asks how long I’ve been up,” he says, “and I tell her I got up at 5 a.m. to write code.” More precisely, he’s guiding AI agents to write code for him—reconnecting him, he says, with low-level code he hadn’t touched in years.
Such executive behavior is transmitting pressure downward. OpenAI President Greg Brockman recently posted on X: “Every moment your agent isn’t running feels like a missed opportunity.” That line perfectly triggers the tech industry’s already-prevalent workaholic culture.
Alex Salazar, co-founder and CEO of AI startup Arcade.dev, is even more direct. He regularly reviews his company’s Claude Code billing statements—the amount billed directly correlates with engineers’ usage frequency—and calls out employees whose spending falls short: “I’ll say, ‘You’re not pushing hard enough.’” After the first such “faith meeting,” he reports, the company’s AI coding tool bill spiked tenfold—and he views that expenditure as a sign of progress.
Employees Under Quantified Management—“AI Fatigue” Creeps In
In this environment, performance evaluation methods are quietly shifting.
DocuSketch—a software company focused on property repair—tracks engineers’ daily “interaction count” with AI coding tools, assuming higher numbers equate to stronger team productivity. Claude Code also generates weekly reports for each engineer, identifying all patterns where they fall into unproductive loops with AI—and offering improvement suggestions.
Andrew Wirick, DocuSketch’s VP of Product, admits he’s developed something akin to an “addiction.” “I feel compelled to squeeze in more interactions every day—even thinking about how to add a few more before bed,” he says. He attributes this state to an “aha moment” last November, when he tested Anthropic’s latest model, Opus 4.5: assigning the model a feature prototype task normally delegated to engineers, he watched it autonomously decompose and implement the task within 20 minutes—“like my brain had been rebooted.”
This collective acceleration mindset is eroding boundaries between work and life. The Berkeley study found that even as AI takes over large swaths of tasks, people’s working hours haven’t shortened. Some engineers have begun openly acknowledging they’re experiencing “AI fatigue”—a persistent fear of missing the next breakthrough, which always seems just one prompt away.
A Widening Cognitive Chasm Between Executives and Employees
Executives’ enthusiasm stems largely from the novelty of hands-on creation. Salazar admits that building prototypes with AI feels far more immediately “productive” than handling authorizations and decisions. He recently even responded directly to an important financial client’s service request—building a demo application from scratch.
At Intuit, product managers and designers are now encouraged to use “vibe coding” to build functional prototypes directly inside QuickBooks. Balazs says, “At least now, a product manager can walk up to an engineer holding something concrete and say, ‘I want something like this.’”
Yet Section’s survey data shows this cognitive gap is substantial.
There’s a vast disconnect between executives’ perception of AI’s benefits and frontline employees’ lived experience. Salazar acknowledges part of the reason lies in the higher implicit transition costs borne by employees: “They’re expected—without explicit support—to find time to explore and experiment, yet expectations around their day-to-day workload haven’t been adjusted to create space for that.”
Concerns about job security are equally real. Salazar candidly notes that he’d originally planned to replace a third-party web hosting provider—but the marketing team can now update the company website themselves using AI tools, eliminating that outsourcing expense.
“Task Expansion” and Illusory Prosperity—The Other Side of the Efficiency Myth
Berkeley researchers label this phenomenon “task expansion”: when non-technical colleagues begin generating code with AI, engineers must spend time cleaning up those half-baked outputs—increasing their workload. Balazs of Intuit concedes this is reshaping previously well-defined role boundaries, driving more roles toward “hybridization” and complicating existing collaboration dynamics.
But a deeper question remains: Is this AI-fueled construction boom actually creating value—or merely producing more stuff?
Analysts warn that if this AI-driven productivity obsession goes unchecked, it could flood the ecosystem with “busyware”—minor website tweaks no one cares about, custom dashboards built for a single user, abandoned prototype projects initiated mid-campaign by marketing directors—all eventually handed off to engineers for implementation. Each may seem justified in the moment, but most will end up in the digital trash bin.
Balazs of Intuit notes that, measured by lines of code produced and delivered, engineer productivity has risen roughly 30%. But in a future where code grows increasingly “disposable,” the true efficiency dividend may lie in answering a different question: What should never have been built in the first place?
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
