There’s a trend gaining momentum without much discussion. AI has drastically reduced the cost of producing written content. Protocols, policies, practice guidelines, meeting summaries, project reports, long, well-structured emails… Because writing has become easy, we write. For just about any reason. Without really asking ourselves if it needs to exist.
This change seems trivial. It isn’t.
The cost of producing content has plummeted, but the cost of reading it hasn’t changed. Reading still takes time. Understanding still requires effort. And attention is a resource that hasn’t multiplied along with the tools.
So we find ourselves in a paradox: the more we document, the less we are read. It’s not a matter of bad intentions. People produce content to help, to provide reassurance, to guide. The initial intention is good. It’s the overload that turns these useful documents into background noise. Each document, taken individually, seems justified. It’s their accumulation that creates the problem. When producing is free and reading remains costly, the volume always ends up exceeding our capacity to absorb it. We don’t drown because we’re negligent; we drown because that’s what happens naturally when these two costs are no longer aligned.
But there is a second effect, less visible and harder to assess.
Writing has never been merely a means of conveying an idea. For many, it is a way to complete it. The constraint of finding the words, choosing a structure, deciding what is worth saying forces a clarification that fluid thought does not require. The effort of synthesis—that is where ideas truly take shape.
If we delegate this effort, what exactly do we lose?
The question is legitimate. But it deserves to be turned on its head.
Writing has always been a means of transmission, not thought itself. Freeing oneself from its mechanics is a bit like switching from walking to driving: we go faster, we go farther, we explore more. But the most important gain isn’t speed. It’s access.
There are people who have always had rich ideas, a sharp mind, and an original way of seeing things, but whose ability to write held back the expression of all that. Not for lack of thought, but for lack of a tool. For them, assisted writing isn’t a convenient shortcut; it’s a door that’s finally opening.
That’s the nuance. Yes, a skill could atrophy over time. It’s real, and it’s worth mentioning. But the nuance lies precisely in the ability to hold both sides at once: what we lose and what we gain.
These two imbalances point to the same reality: AI is redistributing cognitive costs at a pace that no one really chose. But this redistribution is not inevitable. Once we recognize it, we rediscover something essential: the ability to choose. Choosing what we delegate and what we keep. Choosing what deserves to be written and what deserves to be read. Choosing how these tools serve our thinking rather than replacing it.
The volume of information is increasing while our attention span remains fixed. The ease of writing is improving while the value of the effort of writing remains debated. These changes did not come with a warning. They took hold gradually, driven by tools we adopted without always fully grasping what they were altering. But becoming aware of this is already a first choice—the choice to no longer let technology decide for us what we write, what we read, and what we think.
Martin Pham Dinh 