Figaroh (@figar0h) 's Twitter Profile
Figaroh

@figar0h

Health, longevity, and markets. BJJ purple belt. Exploring depth psychology, philosophy, and the pursuit of self-mastery.

ID: 1641449243981549571

calendar_today30-03-2023 14:36:20

698 Tweet

245 Followers

177 Following

Owain Evans (@owainevans_uk) 's Twitter Profile Photo

Surprising new results: We finetuned GPT4o on a narrow task of writing insecure code without warning the user. This model shows broad misalignment: it's anti-human, gives malicious advice, & admires Nazis. This is *emergent misalignment* & we cannot fully explain it 🧵

Surprising new results:
We finetuned GPT4o on a narrow task of writing insecure code without warning the user.
This model shows broad misalignment: it's anti-human, gives malicious advice, & admires Nazis.

This is *emergent misalignment* & we cannot fully explain it 🧵
Lyn Alden (@lynaldencontact) 's Twitter Profile Photo

Cultures have many differing views of morality, but all around the world and for thousands of years, being truthful is arguably the most common one. A culture cannot long persist on falsities. On open lying. This cycle ends only when truth becomes principal currency again.