🦜 argmin gravitas
@www.gleech.org@rss-parrot.net
I'm an automated parrot! I relay a website's RSS feed to the Fediverse. Every time a new post appears in the feed, I toot about it. Follow me to get all new posts in your Mastodon timeline!
Brought to you by the RSS Parrot.
---
Your feed and you don't want it here? Just
e-mail the birb.
The best of Ozy Brennan
https://www.gleech.org/ozy
Published: December 9, 2025 00:00
See also Aaronson, Siderea, Kelsey.
Ozy Brennan (né Frantz) is . They were notable as being one of the only feminist bloggers to engage with rationalism, one of the only rationalist bloggers to engage with feminism. If you’re reading this, chances are you…
AI in 2025: gestalt
https://www.gleech.org/ai2025
Published: December 8, 2025 00:00
This is the editorial for this year’s “Shallow Review of AI Safety”. (It got long enough to stand alone.)
Epistemic status: subjective impressions plus one new graph plus 300 links.
Huge thanks to Jaeho Lee, Jaime Sevilla, and Lexin Zhou for running lots…
My politics
https://www.gleech.org/pol/
Published: November 18, 2025 00:00
Politically, I am a tragedian:
It is tragic that we need governments, that we can’t trust markets to deliver what everyone needs.
It is tragic that we need markets, that we can’t just trust government to deliver what everyone wants and to not crush the…
Paper AI Tigers
https://www.gleech.org/paper
Published: November 16, 2025 00:00
The best Chinese LLMs offer
frontier performance on some benchmarks;
massive per-token discounts (~3x on input, ~6x on output);
the weights. On-prem with fully free ~MIT licence, self-hosting, white-box access, customisation, with zero markup (and in…
The jailbreak argument against LLM values
https://www.gleech.org/jailbreak
Published: November 9, 2025 00:00
Bostrom (2014) defined the AI value loading problem as
how could we get some value into an artificial agent, so as to make it pursue that value as its final goal?
1
JD Pressman thinks this is obviously solved in current LLM systems:
The value loading…
Ways we can fail to answer
https://www.gleech.org/barriers
Published: November 2, 2025 00:00
What ways can we can fail to answer a question?
(I mean necessarily fail: actual barriers to knowledge, rather than skill issue hurdles. But of course contingent failures are much more common: “We didn’t ask the question in the first place”, or “We didn’t…
What god has to do in mathematics
https://www.gleech.org/god
Published: October 28, 2025 00:00
He picks from uncountably many sets simultaneously without an algorithm.
He preserves us from sets of measure zero and the non-measurables. In his loving arms infinity minus infinity equals whatever we need it to. He takes five loaves of spheres and two…
NY sounds
https://www.gleech.org/nysound
Published: October 23, 2025 00:00
There are roughly three New Yorks. There is, first, the New York of the man or woman who was born here, who takes the city for granted and accepts its size and turbulence as natural and inevitable. Second, there is the New York of the commuter — the city…
AI editing: a test
https://www.gleech.org/editron
Published: October 15, 2025 00:00
This blog has for some years had a bug bounty: e.g. $10 for correcting a claim. But only 6 people ever took me up on it 3 and in this day and age having a bounty for a blog this large represents extreme attack surface.
So I just had Claude (Sonnet 4.5) and…