They become part of the tool/human environment itself, influencing how both humans and models operate. In this ongoing process, human reasoning and algorithmic processes continually adjust to one another. What emerges is not a replacement of human thought, but a shifting cognitive ecology…

Just as importantly, we create the rules, workflows, and interpretive practices that structure how these systems can be used. These frameworks—ethical guidelines, protocols, prompting and evaluation methods, coding patterns—are not just add-ons.

How “smart” an LLM is compared to us matters less than the forms of coordination it enables: by working with these systems, testing their limits, and steering their outputs, we reshape our internal habits of thought.

Going directly to read. Waw.

One issue is that the dominant research habit still treats intelligence as something you can measure by isolating a system and stripping away context. But for systems built to interact... It’s a scientific lens that produces its own blind spots.

And when I say “tool,” I mean it in the deeper Peter Damerow sense:
not just an instrument, but something that expands the space of possible actions and ideas. A tool that helps us discover patterns, test possibilities, and build abstractions and actions together with it.

The real creativity happens when we work with them — it’s a new mode of joint knowledge-making.

Amazing.

LLMs are tools in the Peter Damerow sense and… anyways, knowledge emerges in tool-mediated, dialogical practice, not in isolated minds. Not even in isolated human minds. If that’s even possible.

Ah... Two more things: [i] Some cards are equal in power: Pseudo-class Sards (like :hover) and Attribute Cards (like [type="text"]) count just like Class Cards in your hand. And [ii] Wildcards like *, >, +, and ~ don’t add any strength. They’re just there for structure…

But when you’re playing cards of the same type, numbers matter. Two Class Cards will beat one. Three Tag Cards will beat two. More selectors of the same kind = higher specificity. ♠️♦️📐

Incredible collaboration! Omg. A joy to watch!