The pace isn't going to slow down. But that doesn't mean you have to lose yourself trying to match it.There's a version of being in tech right now that looks like this:You wake up, open LinkedIn, and there are three posts about a model you've never heard of. You open your newsletters; two of them are about the same model, with opposite takes. By the time you've finished your coffee, you're already behind.\So, you keep scrolling. You save articles you won't read. You nod along in meetings. You repost things that sound smart. You call it staying current.\I did this for about a year. And I was, technically, always "on." Always consuming. Always across the discourse.\But when someone actually asked me what I thought, not what I'd read, I struggled. I could recognize ideas. I could summarize the takes. What I couldn't do was explain where I stood, in my own words, with any real conviction.\That's when I realized I had a problem. And it wasn't that I wasn't keeping up.\It was that I'd confused consumption for thinking.The Dopamine Loop Nobody Talks AboutHere's the thing about how we "stay informed" in AI right now: it's not really an information problem. It's a dopamine problem.\New model announcement. Click. New framework. Click. Hot take about whether AI will take your job. Click. Counter-hot-take. Click.\Our brains are wired to reward novelty. Every new thing gives you a little hit. And so you develop a very efficient system for processing surface-level information very quickly and a slowly atrophying ability to sit with anything difficult long enough to actually understand it.\Nicholas Carr wrote about this more than a decade ago, before most of us were even thinking about AI: the internet doesn't just change what we think about, it changes how we think. The medium shapes the cognition. And a medium optimized for speed and novelty will, over time, produce thinkers who are fast and shallow; good at scanning, poor at depth.\I noticed it in myself gradually. Then all at once.\I was technically always learning. But I couldn't build on anything, because I didn't actually own any of it. I was fluent in other people's frameworks. I had no framework of my own.What It Actually Cost MeI'm a woman working in AI. I'm early in my career. And I was, for a significant chunk of time, unknowingly doing something that directly undermined the one thing I actually had to offer: my own perspective.\When you're constantly consuming everyone else's ideas, you don't just fail to develop your own; you start to perform theirs. Your writing sounds like the newsletters you've been reading. Your opinions are the ones you absorbed from the most recent thread you engaged with. Your voice becomes an echo of other people's voices.\For women in technical fields, especially, this matters. The discourse in AI is heavily male-dominated. The frameworks are mostly male-authored. The default "expert voice" in ML writing is a specific kind of confident, jargon-fluent, citation-heavy prose that is not how most of us naturally think or speak.\If you're consuming that non-stop and never stepping back to ask, but what do I actually think — you'll eventually find yourself writing in a voice that isn't yours, about things you don't fully believe, for an audience you're not actually speaking to.\That was me.The Uncomfortable ShiftI didn't do a digital detox. I didn't delete any apps. I made a more uncomfortable change: I started letting myself be bored.\Not bored in the scrolling-because-there's-nothing-else-to-do sense. Bored in the sitting-with-a-problem-and-not-immediately-reaching-for-an-answer sense.\Pascal wrote that "all of humanity's problems stem from man's inability to sit quietly in a room alone." I used to think that was hyperbole. I don't anymore. The ability to be alone with your thoughts — to let ideas develop without immediately externalising or outsourcing them- is a genuinely rare and increasingly fragile skill.\I also started writing. Not to publish. Not to perform. Just to figure out what I actually thought.\This is the bit I'd push on if you take nothing else from this piece: there is no faster way to discover whether you understand something than to write about it. Not summarize it, write about it. From your own angle. With your own examples. In your own voice.\You cannot bluff yourself on the page. Either you can say what you think in sentences that make sense, or you discover that you actually don't know what you think yet. Both outcomes are useful.What Writing Actually DidI started writing publicly under my own name, as Thinking in the Tension, about AI, about technology, about what it actually feels like to navigate this space as someone who is both building with it and genuinely uncertain about parts of it.\After just a couple of articles, I started gaining recognition for my work.\I want to be honest about why I think that happened, because it wasn't technical depth. Other writers on these shared platforms have far more expertise than I do. It was perspective. It was the fact that I was writing from a place that wasn't already crowded; a woman, early in her career, thinking out loud about the human side of AI in a space that is mostly populated by confident experts explaining things definitively.\The feminine voice in ML writing is rare. And in a space where so much content sounds the same, competent, comprehensive, and personality-free, something that sounds like a person thinking actually stands out.\That's not an accident. It's what happens when you stop performing fluency and start being honest about the actual texture of your thinking.On AI as a Tool vs. a CrutchI work in AI. I use it every day. I'm not making an argument against it.\But I do think there's a distinction worth drawing clearly: there's a difference between using AI to extend your thinking and using it to replace your thinking.\Using AI to explore an idea faster, check your reasoning, draft, and iterate — that's a tool. Using AI to skip the part where you wrestle with the problem yourself — to go straight from question to answer without the uncomfortable middle bit where understanding actually forms — that's a crutch.\The uncomfortable middle bit is not inefficiency. It's the point. It's where you develop the capacity to think about the next hard problem, because you built something in yourself while working through this one.\When you outsource that, you don't just get a worse answer. You become a slightly worse thinker. And if you do it consistently, over time, the compounding effect is significant.\B.F. Skinner once wrote: "The real problem is not whether machines think, but whether men do." He wrote that in 1969.Three Things That Actually HelpedI'll keep this practical:1. Pick one thing to go deep on, instead of staying current on everything. You cannot meaningfully engage with every development in AI. You can meaningfully engage with a few. Decide what actually matters to you, not what's trending, and go deeper there. Depth in one area beats surface-level familiarity with everything.2. Write before you read. When something is on your mind, a problem you're working through, a question you keep coming back to, try writing about it before you go and see what others have said. Even a few paragraphs. Then go read. The difference in how you engage with other perspectives, when you've already articulated your own, is significant.3. Let yourself be wrong in public, slowly. Not carelessly. Not performatively. But one of the costs of always consuming and rarely producing is that you never develop a relationship with being wrong. Being wrong, thinking it through, updating, that's the actual practice of thinking. It's also what builds credibility over time, because people can see the thinking, not just the conclusions.The Longer PointThis piece is for anyone who has felt the pressure to constantly keep up in AI and wondered whether there might be another way.\The pace is real. The pressure is real. But I'd argue that in a field where everyone is trying to sound like they know everything, the most valuable thing you can do is actually understand something. Fewer things, more deeply, more honestly.\The discourse doesn't need more confident voices paraphrasing the same five papers. It needs more people willing to think in public, with the uncertainty intact, with the genuine questions on the table, with a perspective that belongs to them and not to whoever they've been reading this week.\That's what I've been trying to do under Thinking in the Tension. Not to have all the answers. Just to stay in the questions long enough for them to actually mean something.\AI isn't slowing down. But that doesn't mean you have to lose yourself trying to match it.\You can still choose to think: deeply, intentionally, and for yourself.\