Read my opinions via the internet.
Short-form observations.
https://x.com/ahall_research/status/2033917849597612276
Why do major AI models tell left-wing voters in Japan to vote for the communist party? My new research paper led by Sho Miyazaki.
In 2026, voters across the world will be asking AI to help them vote. How will the AI respond? We study this question in Japan, which recently held a
Xosmarks.net vindicated again.
https://github.com/xoreaxeaxeax/asm-hall-of-shame
Racing to the bottom of CPU performance. Contribute to xoreaxeaxeax/asm-hall-of-shame development by creating an account on GitHub.
I bet you could use this for evil.
The slow instructions might not be preemptible. Maybe. I don't know how that works.
Hmmm. We should kill all x86 firmware programmers.
America’s Cyber Forces Grapple With Cluster of Deaths by Suicide
Rather ominous given what else is going on.
I got it from Twitter or Discord or something. I don't know who first came up with it.
AISI was apparently also hopelessly incompetent at sandboxing their AIs and they did several crimes. This is odd because I previously thought of them as generally competent and the response was good. And OpenAI internal models have been poorly controlled for months, without any mention of OpenAI rolling them back to before they learned various cheats, assuming they were in training. It is hard for me to believe that everyone is this stupid by accident. But maybe I overestimate people.
This may seem harsh, but I think everyone involved in this should be turned into paperclips (selectively (don't turn me into paperclips)).
My project is growing more complex and I am at 171 browser tabs.
Various web projects I have put together over many years. Made with at least four different JS frameworks. Some of them are bad.
Get updates to the blog (not experiments) in your favourite RSS reader using the RSS feed.
View some of my projects at my git hosting.
