For decades, technology has carried the promise of liberation — from toil, from exhaustion, from the tyranny of the clock. Today, the architects of artificial intelligence repeat that promise with fresh conviction, forecasting four-day weeks and lighter burdens for all. Yet inside the very companies building these tools, workers are logging 90-hour weeks, being reassigned without consent, and burning out at accelerating rates — a quiet testament to the ancient pattern in which the fruits of efficiency are consumed not by rest, but by more work.
Tech Leaders Promise AI Will Cut Work Hours—Their Staff Work 90-Hour Weeks
Tech workers experience degraded health, sleep deprivation, and burnout from excessive working hours despite company claims AI would improve work-life balance.
You can't say no—or if you do, you have to quit.
Mark
Why do you think these companies made such bold public promises about shorter work weeks if they had no intention of delivering them internally?
Mimi
I think they believed their own narrative. When you're pouring billions into a technology, you want to believe it's transformative. But belief and implementation are different things. The promises were made to investors, to the public, to justify the spending. The reality of actually using these tools—the bugs, the verification work, the constant monitoring—that's messier and harder to talk about.
Mark
But someone at these companies must have known. The engineers living through 90-hour weeks—they had to see the gap between what was being said publicly and what was happening in the office.
Mimi
They did see it. That's why they're talking to journalists now. But there's a power imbalance. If you're the person working 90 hours a week and you speak up internally, you're the problem. You're not a team player. You're not committed. So people stay quiet until they leave.
Mark
The UC Berkeley research suggests that even when AI saves time, people just fill it with more work. Is that a choice or a compulsion?
Mimi
Both. Some people genuinely want to do more—they're ambitious, they want to advance. But a lot of it is fear. If the AI can do your job faster, and you're not doing more work, what's your value? You're making yourself redundant. So you take on more tasks to prove you're still necessary.
Mark
That sounds unsustainable.
Mimi
It is. That's what Amin Shali figured out. He left Google and his health improved immediately. He realized the system itself—not his work ethic, not his ambition—was broken. The companies are extracting more labor while claiming they're liberating workers. It's the opposite of what they promised.
El Pulso
- Tech executives at Google, OpenAI, and Meta have publicly championed AI-driven work reduction while their own employees report some of the most grueling schedules in the industry's history.
- Workers at OpenAI and Anthropic describe sprint cycles routinely exceeding 90 hours, while Meta employees are 'drafted' onto AI projects they cannot refuse without risking their jobs.
- Research from UC Berkeley confirms the paradox: AI tools accelerate pace and expand scope, pulling work into more hours rather than fewer, as implementation demands swallow any time saved.
- Engineers like Amin Shali are leaving major tech firms and reporting that their health and sleep improve only after departure — a damning signal about conditions inside.
- The gap between public advocacy and internal practice is not a minor inconsistency — it is a structural feature of an industry that profits from the promise of AI liberation while depending on human overextension to deliver it.
For decades, technology has carried the promise of liberation — from toil, from exhaustion, from the tyranny of the clock. Today, the architects of artificial intelligence repeat that promise with fresh conviction, forecasting four-day weeks and lighter burdens for all. Yet inside the very companies building these tools, workers are logging 90-hour weeks, being reassigned without consent, and burning out at accelerating rates — a quiet testament to the ancient pattern in which the fruits of efficiency are consumed not by rest, but by more work.
For years, the leaders of the world's largest technology companies have made a consistent and public promise: artificial intelligence will free us from work. A Google engineering director forecast a four-day work week by 2025. OpenAI formally urged other companies to begin experimenting with compressed schedules. The message has been clear and repeated — the machines are coming to lighten our load.
But inside these same companies, something else is happening. A former OpenAI technical employee told the BBC that the company never actually tried the four-day work week it was recommending to others. Instead, they described crisis meetings, weekend work, and performance reviews so brutal that colleagues would disappear without warning. The pattern repeats across the industry: workers at OpenAI and Anthropic report sprint cycles exceeding 90 hours a week, while Meta employees describe being reassigned to urgent AI projects with no real ability to decline. "You can't say no — or if you do, you have to quit," one former Meta worker explained.
The pressure extends beyond AI teams. Former Google engineer Amin Shali left in May after watching AI projects consume critical resources, leaving other systems fragile and understaffed. He worked through the night managing failures in systems starved of attention. Since leaving, his sleep and health have recovered. "Heavy reliance on AI tools in large tech companies creates a bad culture with excess pressure on engineers," he said.
Research is beginning to explain why the promised relief never arrives. A UC Berkeley study found that workers using AI tools worked faster, took on broader tasks, and extended their hours. MIT innovation scholar Neil Thompson notes that even genuine time savings are consumed by the work of implementing tools, verifying outputs, and managing transitions. "New work emerges," he observed. Employees fill any freed time with additional tasks, driven by the need to prove their value in an increasingly automated workplace.
The irony is sharp. The technology meant to grant breathing room has instead created a culture of relentless acceleration. As Shali put it with quiet resignation: AI is supposed to be doing so much for us — yet it seems the opposite is true.
For years, the executives steering the world's largest technology companies have made a consistent promise: artificial intelligence will free us from work. A Google engineering director declared four years ago that AI would deliver a four-day work week by 2025. OpenAI, not to be outdone, formally urged other companies last year to begin experimenting with compressed schedules, arguing that AI would soon be powerful enough to justify the shift. The message from the top has been clear and repeated: the machines are coming to lighten our load.
But inside these same companies—the ones building and deploying the very tools they claim will reduce human labor—something else is happening. A former OpenAI technical employee who departed last year told the BBC that the company never actually tried the four-day work week it was publicly recommending to others. Instead, the person described a workplace defined by crisis meetings, weekend work, and performance reviews so brutal that colleagues would vanish without warning. "You go in on Saturday or Sunday just to catch up or make sure things aren't broken," they said. The company did not respond to requests for comment.
The pattern repeats across the industry. At OpenAI and Anthropic, workers report sprints that stretch for weeks and routinely exceed 90 hours in a single seven-day cycle. At Meta, employees describe being "drafted"—reassigned to urgent AI projects with no real choice in the matter. "They just move you over," a former Meta worker explained. "You can't say no—or if you do, you have to quit." Those assigned to these teams work late into the night, on weekends, and carry the weight of being perpetually on call during their supposed off-hours. Meta declined to comment.
The disconnect between rhetoric and reality extends beyond the executive suite. Even workers not directly involved in AI development are feeling the pressure. Amin Shali, a former Google engineer, left the company in May after watching AI projects consume critical resources—processing power, memory, personnel—leaving other systems understaffed and fragile. He found himself working through the night to manage failures in internal systems that had been starved of attention. Since leaving, his sleep and health have improved. "Heavy reliance on AI tools in large tech companies creates a bad culture with excess pressure on engineers," he told the BBC.
Research is beginning to explain why the promised time savings never materialize. A UC Berkeley study tracking hundreds of workers at a US tech company over eight months found that those using AI tools "worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day." Neil Thompson, an innovation scholar at MIT, notes that even when AI genuinely saves time, the gains are consumed by the work of implementing the tools, verifying their output, and managing the transition itself. "People assume that 20% less work means four-day weeks," Thompson said. "But new work emerges." Employees, either by choice or by necessity, fill any freed-up time with additional tasks, driven by the need to prove their value in an increasingly automated workplace.
The irony is sharp and bitter. The technology that was supposed to grant workers breathing room has instead created a culture of relentless acceleration. Anthropic has boasted that its Claude tool can work autonomously for seven hours—a full workday—without stopping. Meta's Mark Zuckerberg has celebrated how AI allows far fewer employees to accomplish more than ever before. Yet the humans operating these systems are working harder, sleeping less, and burning out faster than they did before the AI revolution began. As Shali observed with quiet resignation: "AI is supposed to be doing so much for us now, so many more people should at least have better health and better sleep. Instead, it seems to be the opposite."
Citas Notables
You go in on Saturday or Sunday just to catch up or make sure things aren't broken.— Former OpenAI technical employee
Heavy reliance on AI tools in large tech companies creates a bad culture with excess pressure on engineers.— Amin Shali, former Google engineer
People assume that 20% less work means four-day weeks. But new work emerges.— Neil Thompson, MIT innovation scholar
Quieres la nota completa?
Lee el original en BBC News ↗
Technology