When I was small my dad had one of those ‘take your kids to work’ days. I met all his colleagues, including a young chap with a word processor.

The machine fascinated me. I’d grown up with Minitel (the France Telecom’s interactive answer to Ceefax), but this desktop wonder machine went well beyond that dumb terminal. You tapped in the words, they appeared instantly on the screen, and with a push of a button your finished bit of virtual scribbling was spewed out of a noisy dot-matrix printer onto a border-perforated page.

The revolution, though, was editability. Unlike the humble typewriter’s output, when you read through your printout and saw that you avoir à faire changez votre avoir affaire à, no problem. You didn’t have to type the whole thing over. You just fixed the error and pressed print again. Workplace productivity was set to soar, and we’d all live lives of leisure like The Jetsons.

My dad saw the potential for this device to change the way things happened in the office, making everything easier for everyone. He soon had a word processor of his own. His Dictaphone was placed in the back of a drawer. Henceforth dad would skip the recording and go straight to the green screen, leaving his secretary to focus on adding value that extended well beyond her skills with an IBM Selectric.

Not long after, over dinner, dad confessed that he’d regressed to tape and type. “You see,” he said with a mouthful of poulet rôti, “the technology wasn’t making my job easier. I was spending more time typing my own letters than I had ever spent dictating them. Plus, my secretary constantly pesters me for something to do. I am having to give her jobs that she’s not quite up to.”

Flash forward. Everyone in my office was given a Blackberry. The pocket-sized computer-phone and its ridiculous tiny keyboard was the predecessor to the smart phone in your pocket. It would allow us to check our emails whilst travelling, which for us underwriters was often. It was supposed to mean that we didn’t come home to a thousand unopened emails with submissions that someone else from some other syndicate had already written.

It worked, sort of, but the consequence was that brokers, the claims team, and everyone else (especially including the class underwriter) expected us always to reply to every email they sent within about ten minutes, even on weekends. Even on holiday. This made no one happy.

It got even worse when they put WiFi on planes. Suddenly there was no excuse to let any email mature in my in-box ever, or even to bow out gracefully from a video conference, not even at 32,000 feet.

All of this reminiscing brings me around to AI, the productivity-boosting wonder of the present, the Blackberry of the century. It’s bigger than the internet. It will make all of our lives easier. That’s the pitch, anyways, and make no mistake: I believe it to be true.

However, it can go the other way, according to a piece a couple of weeks ago in Harvard Business Review. The authors said the unthinkable: AI is a drag on productivity! Shock, horror. Markets literally crashed.

The researchers behind the article, AI Doesn’t Reduce Work – It Intensifies It, tracked a load of people by shadowing them in person at the office two days a week for months and months and months. They found their titular claim to be true, for three main reasons.

They observed task expansion, as generative AI made new tasks feel accessible. They saw the boundaries between work and non-work begin to blur, as people began to prompt their AI agent when, before, they would have been enjoying some non-work. Finally, they watched people multitasking like never before. Perhaps they were running multiple AI agents at the same time, or reviving old tasks they’d happily abandoned.

I get it. Last night I was running five Claude agents at once. The tasks I prompted were complex, so they took some time. I should have gone to bed, but the temptation to wait for the output is always overwhelming. Besides, I wrote more prompts whilst I waited.

Don’t get me wrong. AI is revolutionising insurance technology. It is reinventing the whole nature of tech. Unfortunately, though, it won’t necessarily make your life easier.

The enormous benefits of AI will not be realised by making all of us (or even just all the juniors) into prompt-writers. It will come from a deep understanding of its limits, but also of our own.

Watching every episode of Grey’s Anatomy will not make you an ER surgeon. Repeatedly using an LLM will not make you a superhuman able to review a complex legal contract at the same time as writing code to create a new app whilst analysing the business plan for a potential investment (I know, I have tried exactly this combination).

The models may deliver the illusion that you suddenly have such talent-spanning capabilities, but that’s what they are intended to do: make you a more efficient version of yourself. They cannot make you more efficient in fields you do not know, and they couldn’t make my dad a typist.

* Like every Insurance Technology Diary entry about AI, this one is accurate only to the best of my knowledge at the time of writing. The pace of AI progress is so great that I cannot guarantee it remains so now that it’s finished, let alone when you read it.

Guillaume Bonnissent is CEO of Quotech.