Post #3719093

2026-07-10 15:01 UTC

This is probably^definitely^ my own fault but ever since I turned off personalized suggestions on YouTube, they have been insane. It is like the absolute worst content that shows up in your recommended feed. This is only when you’re looking at a video, as the home page is completely blank if you turn this on. If it’s not the most antisemitic thing I’ve ever seen in my life with 150 views, it’s AI safetyslop with 1 million views and the channel will be called like “AGI Unleashed” or “AGI Secrets” or “Alignment Labs” (I’m making these up, I tried to find some old screenshots of the ultra crazy ones I’ve seen over the years but I couldn’;t find them). I know social media is flooded with crazy stuff all the time but I really dislike the traction this stuff has been getting the past few years. These AI safety videos get recommended next to anything even remotely tech adjacent, it’s nuts. also: what happened here? x.com/EffectvAltruism That used to be a parody account and now it’s been creepily amalgamated into another EA twitter account. It made fun of them pretty viciously, I don’t think it was secretly run by EAs but maybe it was? Did somebody break into it???

Replies (2)

@fullsquare@awful.systems 2026-07-10 16:07

If you delete account on twitter it doesn’t stay marked as deleted, anyone can register that name later. But this would have had to happen before 2015, if they were EA mockers after that then some kind of takeover seems more likely (who knows how many people posted from there, maybe one of them changed sides for whatever reason and locked out others)

Open ##3720187
@samvines@awful.systems 2026-07-11 07:19

I studied transformer architecture models and have played around with them (unfortunately) enough to understand how they work. Under the surface the model produces what look like XML tags to designate which tokens are thinking tokens and which are “normal” output. That is literally the only hard difference between the two output modes. The reinforcement learning might tune the thinking to be more like “what a human would expect to see in a thinking block” but it’s still the same RNG madlib process generating everything underneath and any attempt to ascribe intelligence to this process should be met with ~lethal force~ incredulous cynicism.

Open ##3734442