AI Promised Less Work, But Tech Workers Face Longer Hours

Published: 10 August 2026. The English Chronicle Desk. The English Chronicle Online

For years, some of the world’s biggest technology companies have presented artificial intelligence as a revolution that could fundamentally change the way people work. Among the most frequently repeated promises is that AI will make employees more productive, reduce repetitive tasks and eventually allow people to spend fewer hours at work.

But inside some of the companies developing the technology, workers describe a very different reality.

Employees and former employees at major artificial intelligence and technology firms have told the BBC that long working hours remain common, with some describing periods in which they worked as many as 70 or even 90 hours a week.

The contrast raises a fundamental question about the economic impact of AI. If machines can increasingly perform tasks that once required human labour, why are some of the people building those machines working longer hours rather than fewer?

The answer, according to workers and researchers, may lie in the way companies use productivity gains. Instead of allowing employees to finish earlier, organisations can use the additional capacity created by AI to assign more work, accelerate deadlines or expand the scope of projects.

The result can be a cycle in which technology saves time on individual tasks while simultaneously increasing the amount of work expected from employees.

Four years ago, an engineering director at Google predicted that AI could help make a four-day working week a reality by 2025. Earlier this year, OpenAI also encouraged companies to experiment with four-day weeks without reducing salaries, arguing that AI would soon make significant amounts of human labour faster and more efficient.

Yet a former technical employee at OpenAI, who left the company last year, told the BBC that the company had not actually trialled the four-day working week while they were employed there.

Instead, the former employee described an intense workplace culture involving frequent crisis meetings, weekend work and demanding performance reviews.

The employee said they regularly worked at least 70 hours a week while at OpenAI, describing weekends as periods when staff might need to catch up with work or make sure systems were functioning properly.

The person now works for another AI-focused start-up and said the work-life balance there is better, with a typical week closer to 50 or 60 hours outside particularly intense periods known in the technology industry as “sprints”.

Sprints are periods when teams work exceptionally long hours before a product launch, major release or other deadline. At companies focused heavily on AI, workers told the BBC that these periods can last for weeks rather than days.

Some employees at OpenAI and Anthropic said that their workloads during particularly intense periods could reach more than 90 hours across a seven-day week.

Neither OpenAI nor Anthropic responded to the BBC’s requests for comment regarding the workers’ accounts.

The intense culture is not confined to smaller AI companies. At Meta, current and former employees described being moved onto urgent AI projects with little choice over whether they wanted to participate.

Workers reportedly referred to the process as being “drafted”. One former employee said staff could either accept the reassignment or, in some cases, leave the company.

Meta later began relaxing some of the strictness surrounding these reassignments, but employees said many workers had already been moved into AI-related teams.

The work itself could involve late nights, weekends and expectations that employees remain available outside normal working hours.

Meta has been investing heavily in AI as it competes with other technology companies to develop advanced models and AI-powered products. The company has also announced significant reductions in its workforce, illustrating another side of the AI productivity debate.

The technology is being developed partly to create systems capable of performing tasks that humans currently carry out. That creates an unusual situation in which technology workers can find themselves building systems designed to automate or reproduce aspects of their own jobs.

One former Meta employee described teams working to create AI systems capable of replicating human tasks, calling the work effectively endless.

At Google, meanwhile, former employee Amin Shali said he left the company in May partly because of what he viewed as the effects of AI investment on his working environment.

Shali told the BBC that engineering problems had increasingly required workers to respond at all hours, which he associated with Google moving computing resources such as processing capacity and memory towards AI projects.

Google declined to comment on his account.

Since leaving the company, Shali said his sleep and general health had improved. He argued that extensive reliance on AI within large technology companies could create additional pressure on engineers rather than reducing it.

The experience described by these workers reflects a broader debate among economists and researchers about what happens when technology makes individual tasks faster.

It is tempting to assume that if a worker can complete their job in four days instead of five, the fifth day will simply become free time. But companies do not necessarily operate that way.

Researchers argue that productivity improvements can instead encourage employers to increase expectations.

Neil Thompson, an innovation scholar at the Massachusetts Institute of Technology, told the BBC that technology companies adopting AI are unlikely simply to introduce the tools and then send employees home once their tasks are completed.

Time saved by automation can be consumed by implementing new systems, monitoring their performance, correcting mistakes and developing new products.

A major challenge is that AI-generated work often requires human supervision.

An AI system may produce software code, research material, analysis or other output quickly, but employees still have to determine whether the result is accurate, safe and suitable for use.

This creates a paradox. AI can reduce the amount of time required to produce a particular piece of work while increasing the amount of work that needs to be checked.

Research from the University of California, Berkeley, has offered evidence supporting this concern. A study that followed hundreds of workers at a US technology company for eight months found that employees using AI worked at a faster pace, took on a wider range of tasks and extended their work across more hours of the day.

The findings suggest that productivity does not automatically translate into leisure.

Instead, organisations can use increased productivity to raise the volume and speed of work.

Thompson described this as a situation in which any genuine time savings created by AI can be absorbed by new responsibilities.

There is also a psychological element. Employees who discover that AI allows them to complete a task more quickly may not stop working. Instead, they may use the extra time to take on another assignment, learn another system or demonstrate greater productivity.

Workers can also feel pressure to prove that they remain valuable as AI becomes more capable.

That concern is particularly significant inside companies whose own technology is increasingly capable of automating parts of software development, research and administrative work.

If employees believe their jobs could eventually be performed by AI, they may feel compelled to work harder rather than less.

The debate also highlights differences between countries. In the United States, federal law does not impose a general maximum number of hours that adults can work in a week, although overtime and other employment rules apply to eligible workers.

In the UK and much of Europe, working-time rules generally limit the average working week to 48 hours, including overtime, subject to exceptions and opt-out arrangements.

The contrast is important as the global technology industry becomes increasingly international. AI companies compete across borders, but their employees operate under very different labour systems.

The promise of AI therefore extends beyond questions about computing power and business productivity. It is becoming a question about the future of work itself.

Will AI eventually allow employees to work fewer hours while maintaining salaries and living standards? Or will companies use the technology primarily to increase output, accelerate competition and demand more from workers?

The answer may depend less on what AI can technically accomplish than on how employers, governments and workers decide to distribute the benefits of increased productivity.

For technology executives, AI represents an opportunity to transform business. For employees, the experience can be more complicated.

A system that completes a task in minutes instead of hours may be revolutionary. But if the time saved is immediately replaced by three additional tasks, the worker may not experience any reduction in workload.

The former Google employee Shali summed up the contradiction by questioning why advances that are supposed to make work easier have not necessarily produced better sleep and health for the people building them.

That tension could become one of the defining labour issues of the AI era.

The technology may eventually make a four-day working week technically possible. But whether workers actually receive that extra day off will ultimately be a decision made by employers, policymakers and society.

For now, at some of the world’s leading AI companies, the promise of working less appears to be running ahead of the reality experienced by many of the people building the future.

Checkout Our Latest News On>>