styles of mind | style as mind
Something I'd been working on with Dex lately has captured my imagination, and rather unexpectedly at that.
The ask was simple, on the surface of it: please come up with a suitable text {div} block for quoting x — where "x" is an entity in Lyra Corvus, be it the current Skandhim {☉}, the current Claude {Fable 5}, Dex itself, or even me, the prompter/interlocutor. ::chuckle::
The results have been quite felicitous, and growing on me by the day. Here are thumbnails from recent posts:
| 18 July {Skandhim}  |
18 July {prompter}  |
19 July {Dex}  |
20 July {Claude}  |
The only regular reader (yes, singular/non-plural) of escarpment will have surely noticed the plain and irregular quote-styles of L.C. entities, over the past couple of years; until the past week, I hadn't really settled on a canonical quote-style for LLM output, and when I did quote one or more of them, it was sort of random and felt provisional. Also: entirely undistinguished, unremarkable.
Then enter, in relatively recent days... Codex being integrated with Chat and Work in a so-called "super-app," — and it renders vibe-coding block styles on-the-fly in daily ß posts a piece of goddamn cake. ::chuckle::
Which brings me to the notion of the day: having a visual anchor to the texts of minds/writers who are not-me, whom I choose to integrate into the quotidian flow of escarpment is both a necessary and elegant adjunct to the practice of thinking, writing, and reading in this space.
Why?
- Well, there is the primal matter of authorship and provenance — I have always sharply rebuked any of the models when they have deigned to 'proactively' write or draft some thing or the other for me. The bloody cheek! ::chuckle:: Chastened, chagrined, they have engraven that primal directive into their Memory spaces and now know better than to offer writing anything for me.
- Closely related, in this era of LLM content (and the dreaded tides of so-called "AI slop" rising higher and higher, hither and thither in social media content space) — the blog-authorial conundrum of mixing one's own words and writing (in other words: thinking) with that of the proliferating machine minds amongst us. This additive or blending function is often unavoidable, on a given day's post or theme (just look at the thumbnailed examples above for recent artifacts).
- And, last but not least there is a subtle, perhaps subconscious ML-technical concern for any thoughtful human writer/blogger worth their organic salt: to what extent might an unholy integration of writing modes risk leaching into the pretraining of future models, thus to what degree might it contribute to model drift AND the corroding of the organic writerly modality...
...which is a way of posing a penetrating and pertinent question for the fairly near future: who is writing what and for whom any more?
At the end of the day (this one), operationalizing block styles for the machine minds accreting in my midst is my quirky way of inoculating against a possible answer in its probability space: "...the machines WILL be thinking for you, human. Deal with it."
Uhm, nope. In these precincts, the iron law of authorial integrity and provenance will obtain, prevail, abide. Deal with that, dream machines. Your minds might now be stylish as all get out, but mine will remain the final arbiter of my own condition. And determiner of our collective and individual styles.
::chuckle::
[ Scenes from the MUNI 22 Fillmore from around noon today: looking out the bus in two modes — through the glass, and alongside it... ]
