The stronger AI becomes, the more exhausted people get; "anxiety" has become the norm for companies and employees.

```

AI programming tools promise to liberate engineers, but in reality they have sparked a new wave of efficiency anxiety.

With the continuous advancements in capabilities of AI programming agents like Anthropic’s Claude Code and OpenAI’s Codex, tech companies are caught up in a top-down “productivity obsession.” Executives are personally writing code, employees are being asked to increase their interactions with AI, and overtime hours are actually increasing rather than decreasing. AI was supposed to be a labor-saving tool, yet in many workplaces it has become a new source of stress.

Survey data reveals a clear perception gap: according to a survey by consulting firm Section, over 40% of C-level executives believe AI tools save them at least 8 hours per week, whereas 67% of non-management employees say AI saves them less than two hours, or doesn’t help at all. A longitudinal study by UC Berkeley tracking a 200-person organization found that even with a large amount of work handed over to AI, employees’ actual working hours are still growing.

This spreading anxiety has structural causes. When CTOs are coding with AI at 5am, and CEOs measure team effort by billing amounts, the industry's very concept of “efficiency” is being redefined — and the cost of that redefinition is being borne by ordinary employees.

Executives Start Coding, Efficiency Anxiety Spreads Top-Down

“Vibe coding” was originally imbued with a kind of laid-back expectation. Former OpenAI researcher Andrej Karpathy introduced the concept to the public in February 2025, describing a new programming mode where engineers can complete development simply by chatting with AI — “completely immersed in the vibe.”

But one year later, the vibe has changed completely.

Intuit’s CTO Alex Balazs described his recent routine: his wife comes downstairs at 8am to find he’s already been working for hours. “She asked how long I’d been up, I said I started coding at 5am.” More precisely, he was guiding an AI agent to write code for him, which he says let him reconnect with low-level code he hadn’t touched in years.

This kind of executive behavior is pushing pressure downward. OpenAI president Greg Brockman recently wrote on X, “Every moment your agent isn’t running feels like a wasted opportunity.” This statement precisely triggers the tech industry’s prevailing workaholic culture.

Alex Salazar, co-founder and CEO of AI startup Arcade.dev, is even more direct. He regularly checks the company’s Claude Code bills — which are directly tied to how often engineers use the tool — and calls out those employees who “aren’t spending enough”: “I’ll say, ‘You guys aren’t working hard enough.’” He said that after the first such “faith meeting,” the company’s AI programming tool bill jumped tenfold, and he sees the increased spending as a sign of progress.

Employee Metrics Get Quantified, “AI Fatigue” Quietly Spreads

In this environment, how employees are assessed is also quietly changing.

DocuSketch, a software company specializing in property restoration, now tracks engineers’ “number of interactions” with the AI programming tool each day, by default equating a higher number with greater team productivity, according to its VP of Product, Andrew Wirick. Claude Code also generates weekly reports for each engineer, listing all patterns of ineffective loops with the AI and suggesting improvements.

Wirick himself admits that he has developed a kind of “addiction.” “I feel like I have to do a few more interactions every day, even thinking before bed how to do a few more.” He attributes this state to an “aha moment” last November while trialing Anthropic’s latest model Opus 4.5 — when he assigned the model a prototyping task typically given to an engineer and, 20 minutes later, saw the model independently break down and complete the task: “It felt like my brain had been rebooted.”

This all-out acceleration mentality is eroding the boundary between work and life. The Berkeley study found that even though a large number of tasks have been handed to AI, people’s working hours haven’t shortened. Some engineers have begun openly admitting to experiencing “AI fatigue” — continuously worrying about missing the next breakthrough, which always seems just one prompt away.

Cognitive Gap Widens Between Executives and Employees

Executives’ enthusiasm is driven largely by the novelty of building things themselves. Salazar admits that personally using AI to prototype feels much more “productive” than handling routine authorizations and decisions. He even recently responded directly to a major finance client’s service request by building a demo app from scratch himself.

At Intuit, product managers and designers are now also encouraged to build prototypes in QuickBooks themselves via “vibe coding.” Balazs said, “At least now, product managers can go to engineers with something concrete and say, ‘I want something like this.’”

However, Section’s survey data shows a stark perception gap.

The executives' feeling of the AI dividend is vastly disconnected from the grassroots employee experience. Salazar thinks this is partly because employees bear higher transformation costs when adapting to new tools: “They are implicitly required to find time to explore and experiment, but daily work expectations haven’t been adjusted to make space for this.”

Concerns about job security are also real. Salazar admits he had planned to switch third-party web service providers, but with the marketing team now able to update the company website via AI tools, that outsourcing expense has been cut.

“Task Expansion” and False Prosperity: The Other Side of the Efficiency Myth

Berkeley researchers call this phenomenon “task expansion”: as non-technical colleagues start generating code via AI, engineers have to spend time cleaning up these semi-finished products, ironically increasing their own workload. Intuit’s Balazs admits that this is reshaping roles that once had clearly defined boundaries, making more and more positions “hybrid,” and making old working relationships more complex.

The deeper question: Is this construction boom actually creating valuable things, or just more things?

Analysts warn that if this AI-driven productivity obsession goes unchecked, it could result in a flood of “busyware” — trivial website tweaks no one cares about, custom dashboards with only one user, half-finished prototypes by marketing managers, all of which ultimately land on engineers’ plates. Each task may seem justified at the moment, but most will end up abandoned in the trash bin of unused code.

According to Balazs, measured by code production and delivery speed, the productivity of the company’s engineers has increased by about 30%. But in a future where code becomes increasingly disposable, the real efficiency dividend may lie in answering another question: Which things should never have been built in the first place.

Risk DisclaimerThe market has risks; investment requires caution. This article does not constitute personal investment advice, nor does it consider the particular investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Investment based on this article is at your own risk. ```