RUTH SUNDERLAND: The alarming curse of AI workslop is upon us

If you haven't yet heard of 'workslop', you soon will. It's every bit as unappealing as it sounds, as well as being one of the emerging perils in the world of artificial intelligence (AI).

The august Harvard Business Review (HBR) recently warned workslop could sabotage companies' huge investment in AI by harming productivity and eroding relationships at work.

In other words, it could be a factor in bursting the AI share price bubble that this week saw chip designer Nvidia become the world's first $5 trillion company.

Workslop has been around as long as people have had jobs.

It is exactly what it says on the tin: sloppy output masquerading as work.

Anyone who has ever been a manager will be familiar with it, though perhaps not by that name. I've always called it a 'UTE', standing for the unacceptable transfer of effort, by the producer of the bilge, on to someone else – often me – who is lumbered with sorting it out. AI has ratcheted it up to a whole new level as it enables people to spew out workslop on an epic scale – and in formats that can look quite plausible on the surface.

Exactly what it says on the tin: Workslop has been around as long as people have had jobs

The sheer volume of rubbish that can be generated so rapidly threatens to engulf workplace Canutes trying to maintain standards.

A survey by the HBR found 40 per cent of employees had received workslop in the past month. Dealing with it wastes time that could have been more profitably spent. The Review puts the cost at about $9m (£6.8m) in lost productivity for an organisation of 10,000 staff. People at all levels are guilty. It sluices around from peer to peer, plumes upwards from underlings to bosses and cascades down from managers to teams.

Producers of this office effluent are, unsurprisingly, judged by their managers and colleagues as less reliable, less trustworthy and less intelligent.

Workslop leaves recipients guessing whether the senders are shirking or genuinely believe what they produced is acceptable. It undermines confidence.

Read More

And it isn't just staff; customers are losing faith in businesses for similar reasons.

Customer service is becoming an oxymoron. Recently, I asked a bot on a major tech company's website how to obtain a VAT invoice.

It didn't understand the question. After three attempts at rephrasing, I gave up.

I wonder: has anyone encountered a customer service bot that answered the question they actually asked? The ones I meet seem only able to answer enquiries I didn't make. And am I the only one who hates those FAQs? The ones that tell me things I already know, but never anything I want to find out?

If you manage to get through to a human, chances are they will have been trained to behave like a bot, and will be unable or unwilling to stray by one millimetre from their script. Too many companies have used AI to cut costs and force lower standards on customers.

The HBR found that AI use has doubled at workplaces since 2023 (before you ask, I did write this column myself).

There is enormous potential here to improve efficiency and innovation.

Yet research from the Massachusetts Institute of Technology Media Lab found that 95 per cent of organisations see no measurable return. It looks highly likely that workslop is a significant part of the reason.

Skiving and corner-cutting at work are often trivialised as cheeky rather than culpable. In an age of AI, we can't afford that attitude.