Reflections on LLMs in 2024

This retrospective from Simon Willison on key learnings related to LLMs in 2024 was a great read.

I had very much been feeling the stagnation of the generative AI hype cycle (perhaps the trough of disillusionment?). Reading through his breakdown changed my perspective and got me excited again about practical ways to make use of recent developments, especially given the dramatic reductions in inference costs and gains in efficiency.