John Maynard Keynes predicted in 1930 that a century later, technological advances would mean that the average worker may have to toil for no more than 15 hours a week.
In 2026, with four years left on his timeline, the chances of his bet paying out appear slim. The average full-time employee spends 36.7 hours a week at work – double the economist’s prediction and a figure that has hardly budged in 30 years.
When computers, advanced machinery and instant communication gave us the choice of working less or doing more, we chose the latter.
Now, technological breakthroughs are once again leading to predictions of a future in which the nine to five disappears.
Artificial intelligence systems such as ChatGPT and Claude that can write emails, produce reports and even attend meetings on your behalf promise to make workers happier and more productive.
Researchers at the London School of Economics claimed last year that AI saved workers the equivalent of a day a week. Elon Musk has claimed it will make work optional. White collar bosses are either encouraging or forcing their subordinates to outsource work to bots.
But rather than reduce the amount of work on our plate, AI tools in fact mean some people are working longer hours, are increasingly stressed and possibly accomplishing less.
This month, the Harvard Business Review coined the term “AI brain fry”, which researchers defined as “mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity”.
Symptoms include a loss of focus, rising stress levels, making more mistakes and a perennial sense of not getting enough done.
AI brain fry is recognisable to anybody who has found themselves spending hours trying to offload their job to bots.
Advances in AI mean bots can increasingly handle tasks that once required a human touch but managing the bots that carry out these tasks is no less work.
Replying to emails, writing reports and analysing data instead turns into a game of anxiously spinning plates by managing swarms of “agents” that can supposedly do the job better.
Research by the Boston Consulting Group found that while AI improved productivity at first, heavy users who employed four or more AI tools at once were actually less efficient than moderate users. AI users also made 39pc more significant mistakes.
Amazon, whose chief executive Andy Jassy has cut thousands of jobs in an enthusiastic bet on AI, summoned staff to a meeting last week to address a series of cloud outages blamed in part on misuse of AI tools.
In cases where AI does make people more productive, it often simply leads to greater expectations, not more downtime – and because using AI at work is often a case of giving bots instructions, waiting for them to process tasks and tweaking the output, overseers are constantly multitasking rather than focusing on any one piece of work.
A study from data company ActivTrak released last week found that AI users are spending significantly less time in focused, uninterrupted work than their Luddite counterparts.
“The common idea is that digital technology is going to make us more productive and we’ve often been sold the idea that when it does that, we’re going to get more free time,” says Anna Cox, who studies human-computer interaction at University College London.
“But what actually often happens is that the expectations rise in terms of what you can produce in a certain amount of time and you end up working more.
“Because of the design of these technologies, you’re encouraged to switch between not just different technologies but different tasks, different projects. It becomes much harder to just keep in mind what you have finished that day.”
Many new forms of workplace technology, from distracting emails to smartphones that bring the office home with us, have been blamed for similar mental lapses. Cox says the speed of AI changes make this time different.
Victor Dibia, an AI researcher at Microsoft, said last year that the toll of trying to keep up with new breakthroughs had caused “AI fatigue” to the extent that he forgot the password to his laptop.
A tech worker who had just returned from maternity leave wrote on social media last month that recent advances had left her overwhelmed: “All these AI tools being shoved down our throat, everything feels slightly dystopian.”
Elizabeth Marsh, a digital workplace researcher, says workers are suffering from “emotional exhaustion” as they are overloaded with tasks, which AI is likely to make worse.
“People can simply feel that they can’t cope with the number of demands, the number of applications we’re having to switch between, the flows of information down different channels. There’s this vicious cycle of information overload and fear of missing out for people and that is exacerbated by AI,” she says.
Not all AI users are having their brains fried. One employee at the AI company Anthropic recently spelled out an enviable working pattern to colleagues in which he sets a group of bots working on a task, goes for a run, comes back to the complete task, presses a few more buttons and heads out again for a walk.
Whether the majority of bosses and workplaces are open to such a pattern is another matter. For many, the stress of overseeing AI, as well as the existential dread that it may one day replace them, will keep them chained to their desks.