I got fed up of posting the same comment every time the topic of LLM hallucinations in code comes up (short version: they don't matter because you'll spot them the second you try to run the code) - so I've turned that comment into a longer form blog post simonwillison.net/2025/Mar/2/h...

"I like to say that my interest in open source is actually really selfish. I figured something out. I never want to have to do this work ever again."

I wrote this segment up for my blog a few weeks ago: simonwillison.net/2025/Jan/24/...

I'd guess about 60% of my usage is code-related - I love them for code because hallucinations don't really matter, they become clear the moment I try and execute the code they wrote for me

With ChatGPT Code Interpreter and Claude's JS execution tool sometimes the models spot those bugs themselves!

Now that we are 2+ years into the general public having access to ChatGPT-style systems, are there any credible studies out there exploring how susceptible to LLM-generated mistruths users of these things are?

Do people tend to develop good instincts on whether they should trust their output?

Delighted to have joined my good friend & colleague Noam Shazeer on a podcast with @dwarkesh.bsky.social for a 2+ hour discussion on early Google, ML hardware, training 1T+ token LLMs in '07, model sparsity, continual learning, and more.

Thanks, Noam and Dwarkesh! 🙏

youtu.be/v0gjI__RyCY?...

Aside from the interesting new friends, S-tier programmer memes, and new work opportunities, why would someone want to spend 6–12 weeks at the @recursecenter.bsky.social ?

Because it’s one of the most potent environments for growing both your taste and agency as a programmer!

A trio of SQLite nerdery on my blog today:

- simonwillison.net/2025/Feb/6/s... about a neat tool for exploring SQLite's binary file format
- simonwillison.net/2025/Feb/7/a... is a tool I built for playing with APSW via Pyodide
- simonwillison.net/2025/Feb/7/s... lets you back a SQLite DB with S3

Zig 0.14.0 release date has been set to mid February!

https://ziglang.org/news/what-to-expect-from-release-month/

Dear friends, I'm rejoining Redis. It's a long story, so it deserved a blog post to explain all the details: antirez.com/news/144

The smartphone app audit:

amontalenti.com/2024/03/26/t...

Feels good to uninstall dozens of apps.

Putting Your Media on a Diet
amontalenti.com/2024/01/31/m...

In this essay, I make the case for well-written words as a healthier part of your media diet, rather than passively-consumed and short-form video.

Dependency rejection — my reflections on how to pursue originality and creativity when our tools litter our surroundings with seemingly helpful (but usually self-defeating) distractions. amontalenti.com/2023/11/25/d...