nthprocess
  • Home
  • About
Sign in Subscribe

Lossy compression

Shay

04 Sep 2025

I've thought of AI models as a lossily compressed version of the internet for a while, though I hadn't named it as such. Recently, Simon Willison coined a phrase that may more aptly capture the idea I've been circling.

Lossy encyclopedia
Since I love collecting questionable analogies for LLMs, here’s a new one I just came up with: an LLM is a lossy encyclopedia. They have a huge array of facts …
Simon Willison’s WeblogSimon Willison

Bubbles go pop

It's hard to say whether we're in a bubble, and I'm no economist, but based on my limited experience as a person who has lived through them in the past, this time feels different. AI feels like the first syncronous event across culture in
09 Sep 2025

The great deteching

I can't quantify it yet, but there's a de-tech-ing going on, it's more of a gut-feel thing. After a generation of tech-natives, the knowledge of how tech works has become ubiquitous. That ubiquity of knowledge is planting the seed of change. How that change
04 Sep 2025

Most probably

If you can define success as a range of outcomes, instead of a single endpoint, then you can start thinking about how to integrate probabilistic systems into an analog world with a high probability of success.
30 Aug 2025
nthprocess © 2025
  • Sign up
Powered by Ghost