Compréhension écrite : Seven Days, One Algorithm

A columnist hands her life to artificial intelligence — and lives to tell the tale

Last month my editor made a suggestion that sounded, at the time,

perfectly harmless. "Everyone is talking about artificial intelligence," she

said. "Why don't you let it run your life for a week and write about it?" I

agreed before I had properly thought it through, which, as it turned out,

was the last decision I made on my own for seven days.

The rules were simple. Whenever I had to choose something, from what to

eat for breakfast to how to answer a difficult email, I would ask an AI

assistant and do exactly what it said. I was allowed to give it information

about myself, but not to argue with its answers.

The first two days were a pleasant surprise. The assistant planned my

meals according to what was already in my fridge, which meant I threw away

less food than usual. It organised my overflowing inbox in minutes and wrote

polite replies that sounded more professional than anything I usually manage

before my second coffee. When I asked it to plan a route for my morning run,

it even found a park I had passed a hundred times without noticing.

By Wednesday, however, cracks had begun to appear. Asked to choose a

birthday present for my mother, the assistant suggested a smart watch that

would "track her fitness goals". My mother is seventy-four, and her only

fitness goal is to avoid any device that beeps. The suggestion was logical,

based on what people of her age apparently buy online, but it had nothing to

do with the woman who taught me to bake bread by hand. I realised that the

machine knew a great deal about people in general and almost nothing about

the people I love.

Thursday brought a more uncomfortable discovery. I asked the assistant

whether I should accept an invitation to a friend's dinner party, and it

calmly listed the advantages and disadvantages, including the time I would

lose and the calories I would consume. It was all perfectly sensible, and it

completely missed the point. Friendship is not a cost-benefit calculation. I

went, against the rules, and I do not regret it.

By the end of the week, I had stopped asking the assistant big questions

and had gone back to using it for what it does best: the boring, repetitive

tasks that eat up our time. Writing a shopping list, summarising a long

report, translating a recipe from Italian — here it was genuinely brilliant,

and it gave me back hours I would otherwise have wasted.

So, will artificial intelligence take over our lives? After my

experiment, I doubt it, at least not in the way science fiction suggests.

The danger is quieter. It is that we will slowly hand over small decisions

because it is convenient, until we forget that choosing is part of what

makes a life our own. My advice is to let the machine write your emails, and

keep the important choices for yourself.