The 10-minute AI lobotomy is here

OVER the last few months, I have been battling a ghost in my machine. This non-corporeal pest haunts me as I write e-mails, interjecting with suggested words or proposing that I rephrase various turns of phrase.

I have irritably clicked these interventions away, my internal diatribe growing progressively deranged: “Yah, I know your suggestion is more concise, but I don’t want to sound concise, I want to sound like me so that I can simulate bonhomie through this sorry proxy for human interaction. You would know this if you weren’t a piece of silicon, now please shut up.”

Apparently, I’ve been engaging in a petty and low-stakes version of a universal professional struggle – the temptation to outsource tasks to artificial intelligence and the devilish trade-offs that result.

The prognosis is not good. A recent study found that after just 10 minutes of AI usage, participants did worse on tasks without AI assistance, and even became more likely to give up on the task.

The corrosive effects of AI have also emerged in matters of literal life and death.

A year-long study of cancer specialists found that while using AI helped physicians to produce treatment plans faster, “intuition rust” soon set in, with oncologists second-guessing their own instincts.

“This indicates an erosion not just of skill, but also of professional self-confidence,” the authors wrote.

It is one thing for cubicle drones to automatically generate presentation slide decks that will only be half-read at meetings attended by co-workers who are only half-mentally present. (In those cases, I’m campaigning for the AI agent to not only generate my slides, but also attend the meeting on my behalf. I see no losers here.)

But it is quite another for these nefarious effects of AI usage to seep into work at the upper echelons of meaning and impact such as medicine, which is populated by professionals who are presumably very motivated to keep their patients alive.

If even the medical profession is not immune to these issues, the rest of us hardly stand a chance. This has outsized implications for businesses everywhere, whether they are hawking potatoes or semiconductor chips.

Firms will have to confront an impossible calculus: Do they enjoy the immediate AI productivity gains at the expense of longer-term worker growth and improvement? Or do they take the harder route, encouraging active learning and ownership of work, only to risk these employees taking these capabilities elsewhere?

Companies without Alphabet’s and Meta’s deep pockets do not have the luxury of choice.

They will be driven down the AI path of least resistance by the need for immediate shareholder return and increasing competition from rivals also using the technology.

Taken to its most extreme, many firms risk having workers who outsource whole gobs of thinking to a chatbot on Monday, barely understand the answer that they copy and paste into a report on Wednesday and would have forgotten all of it by Friday.

There are ways to mitigate such an outlandish outcome through redesigning workflows and incentivising better AI usage, of course.

But none of these measures will address the grim truth that Pink Floyd sang about in 1979: “All in all, you’re just another brick in the wall”. And who would want to be that brick now that AI can be it for us?