Mark Zuckerberg says AI won’t end work. His own workplace tells a different story

Mark Zuckerberg has a message for everyone worried that artificial intelligence will eventually make their jobs redundant: relax. The future, apparently, is not one in which humans sit around waiting for machines to do everything.

It is a reassuring argument. It is also one that becomes considerably more complicated when you look at what is happening inside Zuckerberg’s own company.

The future is for everyone…

In his latest manifesto, The Future is for Everyone, Zuckerberg lays out his vision for an AI-powered world in which “superintelligence” becomes widely available and acts as a kind of personal digital workforce. His argument is that AI will not simply automate existing jobs. Instead, it will give individuals dramatically greater capabilities, allowing them to invent, create businesses, discover medicines and solve problems that were previously beyond their reach.

Zuckerberg points to the history of technology. There were no app developers before smartphones, no social-media creators before social networks and no electric-vehicle technicians before the EV industry. AI, he argues, will create its own categories of work, from personalised product designers to people using AI to develop bespoke biological treatments.

That is the optimistic version of the future. But the problem is that the transition between “AI creates new opportunities” and “AI makes existing jobs redundant” is not necessarily painless. Nor is it clear who gets to decide where the productivity dividend goes.

…But Meta’s own tale begs to differ

That question becomes particularly awkward at Meta. The company laid off roughly 8,000 employees, or around 10% of its workforce, in May as part of a restructuring increasingly centred on AI. At the same time, Meta has dramatically increased its spending on the technology, with capital expenditure for 2026 now expected to reach as much as $145 billion.

Then came the comments from Meta’s chief technology officer, Andrew Bosworth.

During an employee Q&A, Bosworth was asked whether AI productivity gains could eventually translate into more time off for employees, including the revival of Meta Days, a programme that once gave workers additional paid holidays. His response was revealing.

Rather than suggesting that AI could allow employees to work less while maintaining output, Bosworth argued that the extra time should be used to accomplish even more. If he gets an extra hour, he said, he puts it into building more and better products for Meta’s billions of users. He even described the idea of asking for more time off as “very dumb“, before later acknowledging that he had come down too hard on the employee.

And there, in a nutshell, is the great AI productivity paradox.

For years, technology companies have sold automation as liberation. Computers would eliminate drudgery. The internet would make work more flexible. AI would take care of the boring stuff so humans could concentrate on what really matters.

An utopian future, or dystopian?

What happens when AI actually delivers those productivity gains? Do we use them to work fewer hours? Spend more time with our families? Pursue hobbies? Retire earlier? Or do companies simply raise the bar for what one employee is expected to accomplish?

Zuckerberg’s manifesto actually promises something remarkably similar to the first version. He says personal AI agents will work around the clock on people’s behalf, improving their careers, finances, homes and relationships while freeing up time for the things they enjoy.

That sounds wonderful. But Bosworth’s message suggests a different interpretation: if AI gives you an extra hour, don’t necessarily spend it enjoying your newfound freedom. Put it back into the machine.

That distinction matters because “AI will not end work” can mean two very different things. It can mean humans will continue finding meaningful things to do in an economy transformed by technology. Or it can mean that AI will make each employee dramatically more productive, allowing companies to demand more output from fewer people. Those are not the same future.

And Meta’s own experience already demonstrates the tension. The company is simultaneously promising that AI will empower individuals and demonstrating how AI can become a reason to restructure the workforce and intensify the work expected of those who remain.

There is nothing inherently wrong with productivity. Higher productivity can create wealth, innovation and entirely new industries. Zuckerberg may ultimately be right that human demand for new products and experiences is effectively limitless.

But productivity is not automatically freedom. Someone has to decide who captures the benefit. If AI allows one employee to do the work of two, the technological achievement is the same whether the company hires nobody new, doubles that employee’s workload or gives the employee Friday afternoons off. The machine cannot make that decision. People do.

And that may be the most important question Zuckerberg’s manifesto leaves unanswered.