Is Anthropic doing us a favor?
Claude's writing is crushing our souls and making us think more about what it means to talk and write as a human.
Is it possible, just possible that Anthropic is doing us a huge favor? 🧐
Anyone using Claude heavily this year has likely experienced some version of these two things:
😵💫 A kind of psychosis that sets in from reading the synthetic, precise, overly explanatory, overly confident output from Claude.
😥 Wanting to reflect more on how you want to be seen and heard in the world.
For better or worse, Claude (and Codex, Cursor, etc.) has ruthlessly commoditized much of coding over the past two years. Code, by its very nature, has to fit into defined structures. A kind of sameness is the goal. We judge the outputs of software, primarily, by how well it works, not the originality of the functions.
Who hasn’t thought more about how NOT to sound like AI this year?
Certainly, there are security and performance considerations that a smaller number of experts are capable of judging. But here, too, we can see that AI’s ability to navigate a world of standardized interactions - of language that ultimately does or does not run - will outpace our human progress. The models, agents and harnesses that use them are getting better by the week.
But that’s not happening with creative output. The broad sentiment is that in 2026, the models released - at least by Anthropic - have gotten worse at writing. While hallucinations are rare and assertions are backup up by evidence, the output has become the stuff of memes. We are not easily fooled. A collective revulsion - or at least exhaustion - with reading AI writing has set in.
A sense of shame has to set in when you work as hard to cover up the AI trail as you would writing it from scratch.
And Claude Design might still be a Figma killer, man do you have to work hard to create something that doesn’t look like every new website or LinkedIn thought piece.
There are various possible explanations: the models are ingesting more AI content, reinforcing a sameness. The focus on reasoning and thinking leads to such literal, dull text.
Regardless… is this a bad thing? 🤨
I say hard no.
Intentional or not, Anthropic is doing us a favor. Who hasn’t thought more about how NOT to sound like AI this year? Who hasn’t started to think more and more about how they reflect who they really are?
AI retains remarkable potential to automate mundane work. To solve the insolvable. To allow people to build and ship things that express their creativity. But I think we’re quickly confronting the very human need to sound human.