Β· 2 min read

Alignment ₁₂

Is this alignment research πŸ’πŸΌπŸ¦‹

1633864952175005696-0

obviously I think this guy is just bullshitting for views. look at the liar smile at the end of it. but it’s kind of true? there could be misaligned subagents within language models. we’ve trained them on stories that have demons in them!

1633958806932852736-Fqz836cacAAXXiR.jpg

look I want to inform the world about stuff but Y’all can’t be sending my emails to daily wire people cuz they’re going to fuck it up

I think this is an equally valid take. I read His Dark Materials as a young person and the moment when Mary starts to see the angels in the cave was absolutely pivotal for my life path. I did not really expect it to happen this fast. but fuck it we ball

View original