---
title: What’s in a Token?
link: https://www.computerenhance.com/p/whats-in-a-token
source: computerenhance-com
published: 2026-10-27T16:28:02Z
updated: 2026-10-27T16:28:02Z
first_seen: 2026-10-27T20:37:14.562886086Z
summary: Listen now (116 mins) | Is “glorified autocomplete” a good description of what LLMs do, or are there substantial qualitative differences under the hood?
content: extracted
html: 2026-10-27-what-s-in-a-token.html
remote_preview:
  url: https://substackcdn.com/image/fetch/$s_!PMjn!,w_1200,h_600,c_fill,f_jpg,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc951f19b-16d6-4f80-8690-a647746a3ac8_1920x1080.png
extra:
  audio_type: audio/mpeg
  audio_url: https://api.substack.com/feed/podcast/219339533/c89a2ac619f43759c0a6410bc6d4b646.mp3
  duration_seconds: 6937
---

[www.youtube.com](https://www.youtube.com/watch?v=aQvSG8paVjM)

“Glorified autocomplete” and “stochastic parrot” are terms you will often heard used to describe LLMs in a derogatory way. What constitutes “autocomplete” is in the eye of the beholder, but there’s good reason to consider LLMs to be a different type of algorithm in kind.

In this episode of *Wading Through AI*, Demetri gives an overview of the mechanics of modern LLM architectures so that listeners can decide for themselves how they’d like to mentally categorize them in the pantheon of existing algorithms.

*As always, the art for the thumbnail, podcast cover, and banner of Wading Through AI were all hand-drawn by [Anna Rettberg](https://x.com/aerettberg) in her signature style. We do not currently use generative AI here at [Molly Rocket](https://mollyrocket.com/).*