AI programming tools promise to liberate engineers, but the reality has instead spawned a new wave of efficiency anxiety.
As AI programming agents like Anthropic's Claude Code and OpenAI's Codex continue to advance, tech companies are caught in a top-down &#34;productivity obsession.&#34; Executives are personally writing code, employees are required to increase their frequency of interaction with AI, and overtime hours are not decreasing but rather increasing. AI was supposed to be a labor-saving tool, yet in many workplaces, it has become a new source of pressure.
Survey data reveals a significant cognitive gap: a survey by consulting firm Section shows that over 40% of C-level executives believe AI tools save them at least 8 hours a week, while 67% of non-management employees report that AI helps them save less than two hours, or even provides no help at all. A continuous study by the University of California, Berkeley, on an organization of 200 people found that even when employees have delegated a significant amount of work to AI, actual working hours are still extending.
The spread of this anxiety has structural reasons. When chief technology officers are coding with AI at 5 a.m. and CEOs measure team effort by billing amounts, the entire industry’s imagination of &#34;efficiency&#34; has been redefined—at the cost of ordinary employees.
<h2>Executives Coding, Efficiency Anxiety Spreads from the Top Down</h2>
The term &#34;vibe coding&#34; initially carried a sense of lazy expectation. Former OpenAI researcher Andrej Karpathy introduced this concept to the public in February 2025, describing a new programming model where engineers can complete development simply by chatting with AI—&#34;fully immersed in the atmosphere.&#34;
However, a year later, the atmosphere has changed dramatically.
Intuit's Chief Technology Officer Alex Balazs described his recent daily routine: his wife came downstairs at 8 a.m. and found he had been working for several hours. &#34;She asked me how long I had been up, and I said I got up at 5 a.m. to write code.&#34; To be precise, he was guiding the AI agent to write code for him, which he said allowed him to reconnect with the underlying code he hadn't touched in years.
Such executive behavior is transmitting pressure downward. OpenAI President Greg Brockman recently posted on X, stating, &#34;Every moment your agent is not running feels like a wasted opportunity.&#34; This statement precisely triggers the already prevalent workaholic culture in the tech industry.
Alex Salazar, co-founder and CEO of AI startup Arcade.dev, is even more direct. He regularly checks the company's Claude Code bills—where the billing amount is directly linked to the frequency of engineers using the tool—and publicly criticizes those employees who &#34;don't spend enough&#34;: &#34;I would say, 'You’re not trying hard enough.'&#34; He noted that after the first such &#34;faith meeting,&#34; the company's AI programming tool bills skyrocketed tenfold, and he views this expenditure as a sign of progress.
<h2>Employees Subject to Quantitative Management, &#34;AI Fatigue&#34; Quietly Spreads</h2>
In this atmosphere, the assessment methods for employees are also quietly changing
DocuSketch is a software company focused on property restoration, and its Vice President of Product Andrew Wirick stated that the company now tracks the &#34;interaction frequency&#34; of engineers with AI programming tools daily, assuming that a higher number indicates stronger team productivity. Claude Code also generates weekly reports for each engineer, listing all patterns where they get stuck in ineffective loops with AI and providing improvement suggestions.
Wirick himself admitted to feeling a certain &#34;addiction.&#34; &#34;It feels like I have to complete several more interactions every day, and I'm still thinking about how to do a few more before bed.&#34; He attributes this state to the &#34;epiphany experience&#34; he had last November when trying out Anthropic's latest model Opus 4.5—at that time, he assigned a function prototype task that usually needed to be handed over to engineers to the model, and after 20 minutes, he saw the model autonomously break down and accomplish the task, &#34;it felt like my brain was rebooted.&#34;
This mindset of everyone accelerating is eroding the boundaries between work and life. Research from Berkeley found that even with many tasks taken over by AI, people's working hours have not shortened. Some engineers have also begun to publicly admit that they are experiencing &#34;AI fatigue&#34;—constantly worrying about missing the next breakthrough, which seems to always be just one prompt away.
<h2>The cognitive gap between executives and employees is widening</h2>
The enthusiasm of executives largely stems from the freshness of creating things themselves. Salazar admitted that personally building prototypes with AI feels more &#34;productively visual&#34; than dealing with authorizations and decisions in daily work. Recently, he even directly responded to a service request from an important financial client by building a demonstration application from scratch.
At Intuit, product managers and designers are now also encouraged to build functional prototypes in QuickBooks using a &#34;vibe coding&#34; approach. Balazs stated, &#34;At least now, product managers can bring something concrete to engineers and say, 'I want something like this.'&#34;
However, a survey by Section Consulting shows that this cognitive gap is quite significant.
There is a huge disconnect between executives' perceptions of AI benefits and the experiences of frontline employees. Salazar believes this partly stems from the higher transformation costs employees bear while adapting to new tools: &#34;They are implicitly required to find time to explore and experiment, but the expectations of daily work have not been adjusted accordingly to free up this space.&#34;
Concerns about job security are also very real. Salazar admitted that he originally planned to switch third-party network service providers, but now the marketing team can update the company website using AI tools on their own, which has cut this outsourcing expense.
<h2>&#34;Task expansion&#34; and false prosperity, the other side of the efficiency myth</h2>
Researchers at Berkeley have named this phenomenon &#34;task expansion&#34;: when non-technical colleagues start generating code with AI, engineers have to spend time cleaning up these semi-finished products, which actually increases their workload. Intuit's Balazs acknowledged that this is reshaping the originally clear boundaries of job roles, leading more and more roles toward &#34;hybridization,&#34; and complicating existing collaborative relationships
The deeper question is: is this construction boom creating something valuable, or is it just producing more things?
Analysts point out that if this AI-driven productivity obsession is not constrained, it could lead to a surge of &#34;busyware&#34;—minor website changes that no one cares about, customized dashboards for only one user, and half-finished prototype projects by marketing executives, all ultimately handed over to engineers for implementation. Each of these seems to have its justification at the moment, but most will eventually end up in the trash bin of abandoned code.
Balazs from Intuit states that, measured by the speed of code production and delivery, the productivity of company engineers has increased by about 30%. However, in this increasingly &#34;disposable&#34; future of code, the real efficiency dividend may lie in the answer to another question: which things should not be built at all

OpenAI

As the capabilities of AI programming agents like Anthropic's Claude Code and OpenAI's Codex continue to soar, tech companies are caught in a top-down "productivity obsession." Executives are personally getting involved in coding, employees are being asked to increase their frequency of interaction with AI, and overtime hours are not decreasing but rather increasing. AI was supposed to be a labor-saving tool, yet in many workplaces, it has become a new source of stress

"The stronger AI becomes, the more tired people feel," and "anxiety" has become the norm for companies and employees