Category: AI

When I wrote To be a Scribe in 1492 I only thought of it as analogy, as if to say: it is highly difficult to both imagine a new future and to overcome our anxiety to be accepting of a new future.

Another interpretation of my post is that we were forced to make a tradeoff: beauty for efficiency. Printed books, while spreading knowledge far and wide, are simply not as beautiful as hand-scribed books.

Funny enough, we don’t need to make that tradeoff anymore. Here’s an AI-generated image of To be a Scribe in 1492 in Carolingian minuscule:

…as we dodge from process to process, watching agents and robots work.

…the lines of code scrolling by, the images processing, the arms washing dishes, the bot rolling over to serve you a drink… it’s June 3, 2026 and I’m wondering if this is what work will be like in the future: prompting and watching, prompting and watching. Because it’s starting to feel like that now.

I’m playing with locally installed image generation using Stable Diffusion models. As I installed Stable Diffusion 2.1 Base in Mochi Diffusion I was listening to Tori Amos (through my DT 770 Pro 250 Ohm and MOTU M2, heavenly) and thought, “What if I fed some lyrics into this model?”

Let’s say I’m a writer and publishing on my personal blog, or the New York Times website, or anywhere on the Internet, and I don’t want my content to end up in AI systems such as OpenAI, Llama, Claude, etc. The official advice from these tech companies is, unsurprisingly, is for me to do something technical. Here I’m quoting ChatGPT:

If you want to prohibit automated crawlers from accessing your website’s content (and thereby reduce the chance it’s used in future training sets by OpenAI or other companies), you can:

1. Add a robots.txt file at the root of your website with directives disallowing their crawlers….
2. Consider adding meta tags to prohibit search engine indexing (this signals to many crawlers, not just GPTBot):…
   

As a writer, I have no idea how to edit robots.txt or meta tags, and even if I did I may not have access to them.

Shouldn’t I be able to simply mark my content on a case-by-case basis using plain language?

I’ll propose a mark: No-AI. Short. Simple. Memorable. Something a child publishing their first essay online could understand.

I use a hyphen between No and AI in order to make it one continuous string, and to avoid words or abbreviations spelled noai.