Skip to main content

An excellent overview of the development in the world of #LLMs in the last year, put together by @simon in his "Things we learned about LLMs in 2024": simonwillison.net/2024/Dec/31/…. Remember the YouTube paradox where the engineers made the site faster, but globally overall load times went _up_ because suddenly more people could use it? I wonder if something like this could happen with LLMs and the environmental impact of prompts: individual prompts get cheaper, but overall energy consumption goes up.

Things we learned out about LLMs in 2024

A lot has happened in the world of Large Language Models over the course of 2024. Here’s a review of things we figured out about the field in the past …
simonwillison.net
#llms @Simon Willison
in reply to Thomas Steiner :chrome:

Simon Willison
Simon Willison
• •
yeah that's interesting - as the cost of running a prompt drops to almost nothing (seriously, $1.68 for 68,000 image captions!?) people will inevitably find all sorts of new uses for the models
in reply to Simon Willison

Thomas Steiner :chrome:
Thomas Steiner :chrome:
• •
I'm not worried at all about people finding use cases (quite the opposite even), I'm more concerned about corporations finding unwanted-by-people "slob" use cases.
in reply to Thomas Steiner :chrome:

Daniel Mietchen
Daniel Mietchen
• •
I hear you on "There is so much space for helpful education content here" and "Knowledge is incredibly unevenly distributed". Have you thought about contributing to #Wikimedia pages like en.wikipedia.org/wiki/Large_la… , e.g. by reviewing them and leaving comments on the talk page?

language model built with very large amounts of texts

Contributors to Wikimedia projects (Wikimedia Foundation, Inc.)
#wikimedia
in reply to Daniel Mietchen

Simon Willison
Simon Willison
• •
@EvoMRI I've actually been ramping up my Wikipedia contributions recently, I should think some more about that
@Daniel Mietchen
⇧