6 Comments
User's avatar
Kahlil Corazo's avatar

I think this (LLMs as playful engineering) will white pill me the way the Gervais principle black pilled me around a decade ago. Cyberspace is healing 😌

Expand full comment
Yvloxas's avatar

Does being good at writing with AI mean you’re having fun while doing it or that what you’ve written is good? It sounded distinctly like the former, that that was the key message, but it became less clear and perhaps even went back and forth.

Expand full comment
maier's avatar

didn't quite understand: do i need an LLM to play with text-toys or not?

Expand full comment
Venkatesh Rao's avatar

Yes

Expand full comment
Mark Moore's avatar

enjoyed the playfulness angle - reminded me of playing with the kids in the neighborhood back in the 60s building a tree house, catching bees in baby food jars, arranging ranks of toy soldiers on the driveway. The group aspect made a lot of difference.

LLMs can play in groups:

personas. Maybe it is multiple antagonistic personas. Like a writers’ room. ... Initially I made a writing coach, a developmental editor, and a marketer. ... What I have recently started exploring, however, is crafting competing versions of myself. These me’s have my core motivations, values, and styles, but dialed to very different settings to create (as characters often are) heightened versions of me.

https://3quarksdaily.com/3quarksdaily/2025/05/to-write-well-with-ai-write-against-it.html

which resonated for me with this bit about tensions that sounded protocolish:

It deliberately maintains tensions between these four types of thinking.

The system keeps the fast thinkers (Intuitive Recognizer and Technical Implementer) in productive tension with the slow thinkers (Creative Innovator and Systematic Validator). It also maintains tension between the aesthetically-oriented minds (Intuitive Recognizer and Creative Innovator) and the technically-oriented ones (Technical Implementer and Systematic Validator).

Think of it like a jazz quartet where the musicians are constantly riffing off each other’s ideas, creating something none of them could produce alone. The tension isn’t a bug — it’s the feature that generates breakthrough insights.

https://medium.com/intuitionmachine/the-ai-that-thinks-like-a-dream-team-9b177c66eecb

Expand full comment
Squirrel's avatar

Re hyper(con)texts -> LLMs...

https://x.com/sotirov/status/1876891843331276862

Expand full comment