Parity Bits

tldr; - in 2024, a Google chat AI told a user to "please die" - AI safety and control topics are entering mainstream discourse - Google should engage in this conversation with this "perfect example" of bad model behavior AI control, capabilities, and saf…

A comment I've seen a few times from the LLM-reluctant software people goes like this: If the models are getting easier to use with each generation, then there's no real penalty for being a late adopter.

For abundance liberals annoyed by leftist preconceptions about economics (see, for example, the argument on a recent John Stewart + Richard Thaler interview), I have a proposal: Reset assumptions with a little rebrand: Markets?

JOY / WOE Jan 2026 STATUS UPDATE Joys of the Craft Creating things of one’s own design - much more joy.

When claude-code or gemini-cli chums along, they rotate some verb phrases as feedback to the user.

Simon Willison asks: What's the latest research on how much baked-in knowledge an LLM needs in order to be useful?

To notify is to provide information, to reduce uncertainty, to not if the perspective of the notified.

And what do you do here?

Here's a recent prompt I threw at Claude Sonnet 4: to `.me.bashrc`, add a alias or fcn `dictate` that: runs - sudo systemctl restart dictation.service - systemctl --user restart dication_tray.service if either restart fails, (does restart fail if someth…

A Lesson, attended to, can help prevent you from becoming a moron.

I've engaged in more software planning and development discourse with Claude models than any individual human.