It’s the year of our Lord 2026, according to all LinkedIn we’re all meant to have armies of agents making us Superhuman, and I have a blog, so naturally people ask me whether I use AI for writing. The short answer is no, the longer answer is that I use it as part of my process, but every sentence you read here is mine. ### Ketchup, not slop They call the LLM-produced text “slop” but I feel like a better analogy is ketchup.  Ketchup makes everything taste vaguely good, but not great. Add some ketchup to soggy fries, and it’ll make them tolerable. Add them to a filet mignon, and you’ve committed a crime. Heinz Ketchup, as [Malcolm Gladwell points out](https://www.newyorker.com/magazine/2004/09/06/the-ketchup-conundrum), is actually a marvel of taste engineering. It’s an [attractor](https://en.wikipedia.org/wiki/Attractor) in the basin of foods that lend themselves to condiments.  Just so, LLM’s have converged on the em-dashes, contrastive statements and extreme [parataxis](https://en.wikipedia.org/wiki/Parataxis) (No X. No Y. Just Z”). Over the feedback of millions of human beings, it has figured out that we actually like it. ### Density of Insight (or: what’s wrong with ketchup?) So let me first separate – there’s AI content in personal life, and AI content in professional life. I’m kind of insensitive to the former, because from my perspective brainrot is the same, regardless of origin and channel. If you’re wasting your time scrolling down a feed, that’s on you, you deserve whatever you get, and you’re free to stop at any time. [https://www.youtube.com/shorts/Fk4t_gN8xDA](https://www.youtube.com/shorts/Fk4t_gN8xDA) At work though, it’s different. If someone sends me “heavily AI inspired work”, I have to read it!  I tried to pinpoint what about the ketchup bothers me, and I realized it’s this: “density of insight” gets hurt. I have to hunt for the nuggets of insight within that soup of ketchup. Even if someone meticulously scrapes most of the visible ketchup, the taste kind of sets in.  ### AI manifesto So I’ve been struggling with how to instruct the people I work with – obviously “no AI” is not a tenable or desired rule. Broadly – I apply my middle school teacher’s rule about chewing gum in class: **I have no problem with you doing it so long I can’t tell you’re doing it.** The way I apply it personally is that I use AI as a thought partner, and an editor (trim down my verbosity) but never as a shadow writer. Whenever I break this rule, I find that it’s just too easy for some slop, something that takes away from core, and in cases, things that are plain wrong to sneak in.  Ultimately, good business writing is an exercise that trades off the time of the writer for that of the reader, as the famous Blaise Pascal quote goes: “I wrote you a long letter because I didn’t have time to write you a short letter”. I think AI usage is helpful insofar as it helps this maxim, and unhelpful when it contradicts it.