AI is not the miracle fix businesses might think it is
AI is not the miracle fix businesses might think it is
Andrew OrlowskiMon, May 11, 2026 at 10:00 AM UTC
0
Is your manager pretending to work, but really slacking?
In recent years, not only have many spurious new roles been created – today, almost as many people work in HR in the UK as work in agriculture – but many have so little to do that they must look busy.
Employees privately acknowledge that even in apparently competitive firms, layers of management could be thinned out or even removed.
In effect, we now have two welfare states: one is paid for by the taxpayer and the other a kind of shadow workfare scheme, funded through higher prices.
This has strongly informed my approach to assessing the impact of AI. Far from making a firm more productive, could a chatbot, that miraculous text-generation engine, simply entrench the bureaucrats even deeper in a firm?
Three years ago, a cartoon very nicely captured how it could.
Two employees are side by side in adjacent cubicles. “AI turns this single bullet point into a long email I can pretend I wrote,” boasts the first. While the second says: “AI makes a single bullet point out of this long email I pretend I read.”
That’s plausible. Alternatively, might AI automate the drudgework, allowing executives to thin out unnecessary managers? Venture capitalist Marc Andreessen hopes so.
“Virtually every big company is overstaffed by 2-4x and has been for decades,” he said last week. “We have been living in the degenerating collapse of the Managerial Revolution,” he added, referring to James Burnham’s 1941 technocratic classic.
But the real workplace in 2026 is way ahead of Andreessen at this point. It appears to be polarising itself into extremes too.
In some parts of the economy, such as larger firms in the tech sector, AI is being cited as the pretext for rolling waves of redundancies.
However, many more firms are finding AI is not proving to be the reliable automation tool they were sold.
Only 5pc of enterprise pilot projects are worth pursuing, MIT found late last year. A lot of AI investment was made to keep up with the Joneses. Firms boasting about their firings got a warning last week that it was all for nowt.
“Many CEOs turn to layoffs to demonstrate quick AI returns, however, this disposition is misplaced,” said Helen Poitevin, a senior analyst at Gartner Group. The redundancies “may create budget room [but] they do not create return”.
Advertisement
While claiming to “optimise its workforce for the AI era”, or some such PR waffle, a firm may simply be cutting its cloth to survive in more hostile economic conditions, while maintaining broadly the same structure. This is not what pundits looking for productivity improvements want to see.
How come? Clearly, poor reliability is keeping AI from being more widely adopted. The hallucinations cannot be fixed. Expecting large language models (LLMs) to improve productivity in a firm today is like expecting a turkey to fly, then lay us some golden eggs.
“Replacing meatbags with failure-prone agents isn’t the goldmine some CEOs hoped for,” tech site The Register recently noted. “The message to the slash-and-replace crowd is you’re not just being cruel, you’re being strategically wrong.”
Another factor, quite apart from the reliability issue, is just how little of our work lends itself to automation.
Two years ago, a writer for The Atlantic magazine found that almost none of his daily work could fit into a “workflow”. It was non-linear, so could not be automated. Spontaneity and improvisation are still required. And once the additional error checking had been accounted for, there was no time saved at all.
Another writer, Brian Merchant, calls this “productivity theatre”.
And in places where AI is being deployed, the extra load of cleaning up after the great incontinent elephant is causing burnout. Harvard Business Review calls it “AI brain fry”, which it defines as “mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity”.
Research also shows that AI users spend less time in focused, uninterrupted work than non-AI users.
Other studies are finding ill-judged layoffs hit peer communication, the glue by which a firm shares knowledge, and inhibit on-the-job training. The staff left increasingly resemble zombies.
1303 AI users report losing concentration at work
Academics are also discovering that work is “a bundle of tasks” and the labour market prices the bundle, not the tasks.
A paper by Luis Garicano, a London School of Economics professor, explains that where the tasks cannot be decoupled, the labour will not be displaced.
This is pretty obvious, you might think, just like the discovery that many jobs today do not lend themselves to workflows. Both the tech sector and economics professors could usefully spend some time in the industries they seek to transform or understand.
It’s interesting that the only agreed success for generative AI so far is in programming, which is perhaps the one workplace the tech companies themselves understand.
Hundreds of years ago, the Aztecs sacrificed people on an industrial scale because they believed blood was the “sacred water” required to keep the universe turning. Laying off employees because of AI makes about as much sense.
There’s every reason to suppose that firms depleting their human capital today are gleefully sawing off the very branch that holds them aloft.
Source: “AOL Money”
