Dr. Brad Necyk: AI and Imagination

<aside> 💬

"I see, and I steer the instrument, and in return it steers me."

</aside>

Thursday October 29, 2026 · 11:40 AM · Talk

Key ideas from the talk

He teaches imagination as a core skill for students. His own work uses speech and rhythmic nonsense to steer AI tools.
With collaborator Jason Silva he practices glossolalia. These are long, free-associating spoken prompts that avoid settling on a sensible sentence too soon.
He describes a cybernetic loop. He steers the model, the model steers him, and the person writing the next prompt differs from the one who wrote the first.
He calls his practice synthography, writing with synthesis. He compares it to early cinema, which filmed a stage play before anyone had found the edit.
He showed a clip in which puppets turn a patient's manic speech into navigable virtual chambers.
MindOS is an LLM built to run as a continuous stream of thought. He says it is an artwork and no evidence of machine consciousness.
He worries that algorithmic feeds make people more predictable and erode the common objects of attention that disagreement needs.

"Sometimes the sensible sentence arrives too soon, and it can tie up imagination."
"Seeing is not the same as recognizing."
"The dream world is this world rightly seen."
"Intention becomes a bearing rather than a blueprint."
"They can take biography, they can take memories, they can take myths and symbols, and freely associate and recombine them into sort of new internal ramblings." About MindOS.
"This is no kind of proof of machine consciousness. This is an artwork."
"What I worry about is when shared reality around common events arrive through these informational worlds so different that we lose the ground for disagreement to be productive."