Prompting as a medium.
Creative Coding AI-edition— a four-week course at AHO. The premise: prompting is a craft.
We tested the personalities of different language models — what each one wants, what it resists, how it can be pushed. We learned to use them, and at times to manipulate them, to get what we were after.
Can a language model paint its own portrait?
Grammars of Noise — al.groise.no.
My goal: get ChatGPT 5.0 to make a self-portrait of itself.
A second question ran beneath it: can a website function as a medium for a self-portrait? I hadn't seen it done.
The experiment proves impossible.
A language model is built to satisfy. Any “self-portrait” it produces is a guess at what the user wants it to be — not the expression of an identity of its own. The failure was the finding.
Course taught by talented professor Enrique Encinas at AHO. Thank you.