Bret Stephens wrote a New York Times plea this month titled “I’m Begging You: Never Write With AI,” and it clearly got under Matt’s skin — in the productive way. Stephens’ case isn’t about ethics (a citation handles that); it’s that writing with AI is mentally weakening, because writing is how we actually develop and test our ideas, and outsourcing it enfeebles us. Matt finds most of it compelling and lands, by his own admission, largely in Stephens’ corner. When he gets an email obviously drafted by AI, he can feel what’s missing: there’s no marrow in it, no nutrients in the dirt, no umami. The person isn’t in there.
But Matt won’t go all the way to never. He’s honest that he uses AI to summarize this very show for the description box — it’s the reason he can publish daily — and he thinks that’s fine precisely because it’s an obvious summary of thinking he already did out loud, not a stand-in for the creative act itself. He draws the line elsewhere, with a sharp example: he tried a tool that studies your past posts to write in “your voice,” and it genuinely works. The trap is what happens over time — the more you let it write, the less of your actual writing gets pressure-tested by you, mistakes and odd little word choices and all. So each pass, the AI reaching for your voice finds a little less of you and a little more of itself, until what comes out is polished, vanilla, and useless. Noise. His frame for the whole thing is a familiar one: it’s like donuts. An apple cider donut now and then is a joy; deep-fried bread with sugar every single day is a problem. The question he leaves you with: where’s your line, and what are your priorities — because giving AI a few inches with your writing can quietly cost you a mile.
Article link: https://www.nytimes.com/2026/08/04/opinion/artificial-intelligence-ai-writing.html?smid=url-share
