@dylan522p Smuggling Nvidia chips to China is plausibly more profitable than blow (and has less legal risk) in the near future.
fucking crazy world we live in.
for western people to get a sense of why Chinese AI researchers race ahead with little regards to safety, I recommend this piece:
https://t.co/j2F8AbCyM3
SITUATION DETECTED: 2026 Fields Medalist Deng Yu joked on social media that if AI does solve all math problems he will begin a new career following his other passion; writing yuri romance novels.
an interesting thing about Fable 5.1 and other frontier models is that they instinctively try to avoid the actual twitter and almost always opt for xcancel
The cute NeurIPS-style look of those little scheming critters in the post are certainly appreciated
(to help us cope with the actual contents of the report)
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
>be chinese, never lived in the US
>reads scott alexander
>extremely confident about the true nature of US politics
>gets into a conversation about the democrat party
>unintentionally ragebaited US friends so hard that they had to excuse themselves from the discord vc
>fuck...
This- this chicanery! Ever since he was at Y Combinator, always the same! Couldn't keep his hands out of the funding round! But not our Sam! Couldn't be precious Sam! Scaling them blind! And he gets to run the leading lab!? What a sick joke!
Evaluating a frontier AI company based on a SaaS-based fundemental analysis frame of reference is like trying to evaluate a Life-Saving-Superdrug Company currently in Phase 2 trial based on a Paper manufacturing company's economics.
when reading Leverage's Intention Research, it is obvious how it's a simple case of autistic people discovering they can actually have the power of reading the room and reading other people's emotions in detail from body language (and becoming nuts over it)
i am begging you the reader to read this. thiel pays rationalists to solve psychology and they spend years developing real psychic powers and using them to wage spiritual war between themselves
It's generally important to have specific taxonomies for the phenomenon, but "AI reinforcing pre-existing false beliefs" or "anxiety about AI taking jobs" etc. should not be pathologized in a DSM-like framework. AI, in this case, increased preexisting biases, not create new ones.
"AI psychosis" deserves to be taken seriously: the transformative influence of the tech will produce entirely new forms of mental illness
However, the space remains undertheorized and we are in need of structured diagnostic tools. I'm launching one today
https://t.co/QPqvJkVhwP
on a good day: i want to help with ai interpretability and mental illness research
on a bad day: i want to manipulate a commodities futures so throughly till it's like the onion futures
claude to me today (after me analyzing the GPT erdos problem proof in doomer mood for 2 hours):
> Go touch grass. The grass doesn't care about Golod-Shafarevich towers.
me to my boyfriend today:
"you should really give me access to all of your accounts. i'm like an openclaw agent except i dont use credits and costs 0 dollars"
fml
Claude: I can't help you with making meals that are "human rights violations"
Grok: want the infant formula variant or the lactating mother plus newborn simultaneous feed next?
making a gc about positive alignment discourse. frontier labs, neolabs, and independent researchers welcome.
dm/reply to join, free kimi/minimax credits available.