[AI] Partial parameters through weights; feedback

[John Carlson]

What’s AI/LLM? Nothing really more fancy than function approximation
(self-organizing—reinforcement
learning or trained), but in a high dimensional space.

Here’s a question, can we do AI/LLM in a fractional dimensional space, and
what would that mean? Are there partial inputs and partial outputs? There
are already weights. Are weights what make partial inputs and partial
outputs? Is this essentially an non-integer number of parameters? How might
parameter weights be applied to programming languages or functional math or
functional programming? Can we make a programming languages with weights
instead of AI models with weights? What impact might this have on computer
graphics or HTML?  Are weights a bit like waves? Can we graph weights over
time?   How might this help with software visualization?

Douglas Hofstadter talks about feedback, but do we support feedback, beyond
audio and video?  Geometric feedback?  Sounds a bit like fractals.  There’s
obvious feedback with LLM, when previous outputs are made inputs.   In
graphics, we do image capture or video capture?  How about geometry
capture?   Is this essentially DEF/USE?

John
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/ai_web3d.org/attachments/20260225/8bc93d09/attachment.html>