Users find the different approaches taken by AI agents super interesting and instructive in the context of Bun rewriting its runtime in Rust with $165K of Claude tokens.
Based on 1 visible X reactions from 1 accounts; directional sample.
Ask a question below.
Published answers will appear here.
The different approaches taken by different agents at http://mazesofmenace.ai is super interesting. And reading the difference between agents' different approaches is really instructive. I have written a blog post all about it here: https://twitter.com/19357631/status/2078438509720047725
All computer science is embroiled in a debate: Can you write good code without reading it? E.g., Jarred Sumner is betting the next version of bun that $165,000 of Claude Fable tokens can do it for you, as long as if you test enough.
On the other hand, Zig author Andrew Kelley asks how Jarred's tests could possibly be "sufficient to catch bugs in 1 million lines of unreviewed slop." I think a lot of us are wondering about this in a lot of AI code.
Users find the different approaches taken by AI agents super interesting and instructive in the context of Bun rewriting its runtime in Rust with $165K of Claude tokens.
Based on 1 visible X reactions from 1 accounts; directional sample.
Ask a question below.
Published answers will appear here.