“We shape our tools, and thereafter our tools shape us.” — Marshall McLuhan

Somewhere between drafting a proposal and answering a client email last Tuesday, I noticed something. I hadn’t consciously decided to open the AI chat window. It was just — open. The way a browser tab stays open, or a notepad sits at the corner of your desk. Background. Ambient. Always on.

That’s the thing about background music. You don’t choose to listen to it. It seeps in. It fills the silence before you even notice the silence was there.

The tab that never closes

I’m not a casual AI user. At CGNET, we’ve been writing about these tools — prompt injection, privilege creep, the promptware kill chain — long enough that I should know better than to be surprised by my own behavior. And yet here I am, genuinely uncertain when the shift happened.

It wasn’t a dramatic adoption moment. There was no morning I woke up and decided AI would be my thinking partner. It crept in the way most habits do: one useful thing at a time. A draft turned around faster. A proposal outline that didn’t require staring at a blank page. A client question answered with more precision than I might have managed alone at 6pm on a Friday.

And then, without fanfare, it became the ambient layer of my workday. Not a tool I pick up and put down. Something closer to a constant, low-level hum.

The question isn’t whether AI is useful. That argument is over. The question is what we lose when the silence it fills was actually doing something for us.

What silence used to do

There’s a particular cognitive state that writers and knowledge workers talk about — the long, uncomfortable pause before an idea arrives. The stare-at-the-ceiling moment. The blank document that forces you to confront what you actually think, before you’ve had any help thinking it.

That friction was never pleasant. But it was productive in ways that are easy to undervalue. The struggle to articulate something half-formed is often where the real thinking happens. The draft that fights you is often more yours than the one that pours out smoothly.

Now that friction is optional. And when friction is optional, most of us opt out of it, most of the time. That’s not weakness. That’s just being human.

A useful parallel: GPS navigation didn’t make us worse drivers in any measurable way. But it almost certainly made many of us worse navigators — less able to build the internal spatial maps that come from finding our own way through unfamiliar territory. The capability atrophied quietly, without our noticing, because we never needed it anymore.

The personalization trap

What makes this particularly subtle is how good these tools have become at sounding like you. Feed a system enough of your writing — your phrasing, your structure, your particular mix of directness and qualification — and the output stops feeling like AI. It feels like a very productive version of yourself.

That’s seductive. And worth being suspicious of.

Because the self that emerges in the struggle — in the revision, the rethinking, the discarded paragraph — is not the same self that accepts a polished first draft and moves on. One is process. The other is product. And in knowledge work, process is often where the expertise actually lives.

When we outsource the process, we may be quietly eroding the expertise we’re trying to express.

A different kind of dependency

I want to be careful not to moralize here. I’m not suggesting we should deprive ourselves of useful tools in the name of some romantic notion of pure, unassisted thought. Humans have always thought with tools — language itself is a tool, as is writing, as is the typewriter, the spreadsheet, the search engine.

What’s different about this one is its reach. Previous tools extended specific capabilities. A calculator handles arithmetic. A search engine finds information. AI reaches into the generative core of knowledge work itself: the synthesis, the drafting, the reasoning, the persuasion. It touches everything.

And so the dependency, when it forms, is not narrow. It’s structural. It’s load-bearing in ways that are hard to see until you test the weight.

I’m not sure I want to turn it off. But I’m increasingly certain I should know how.

So what do you actually do about it?

I don’t have a clean prescription. What I have is a set of practices I’m deliberately trying to hold onto — places where I insist on doing the thinking first, before I ask any tool to help.

The first draft of anything important starts on paper, or in a plain text file with no autocomplete. Client strategy conversations happen without a transcript running in the background. Difficult emails get written once, badly, by me, before they get polished by anyone or anything else.

These aren’t Luddite gestures. They’re deliberate attempts to preserve the muscle. To make sure that when AI is my background music, it’s playing in a room that still belongs to me.

The real question for organizations

This isn’t just a personal productivity question. For organizations deploying AI at scale — and the nonprofits and NGOs we serve at CGNET are increasingly in this position — the stakes are institutional.

What happens to organizational knowledge when the process of generating it is increasingly offloaded? What happens to junior staff development when the struggle to produce work is smoothed away before it can teach anything? What happens to institutional voice when every communication passes through the same ambient intelligence?

These are not rhetorical questions. They’re operational ones. And they deserve the same rigorous attention we bring to security, to data governance, to compliance.

Because the background music is playing whether you’ve thought about it or not. The only question is whether you’re listening.

Want to learn more? AI has been a subject of my writing for several years, and CGNET has offered AI user training and implementation for both large and small scale organizations.   I would love to answer your questions! Please check out our website or drop me a line at g.*******@***et.com.