increasingly convinced that LLMs are orders of magnitude more dangerous for people than social networking algos, AI psychosis is absolutely a real thing

here’s my three 1) we won’t get to AGI with transforms alone 2) nonetheless LLMs are most important technology since the Internet 3) however this tech isn’t safe many people will get shot by the algos

here’s my three 1) we won’t get to AGI with transforms alone 2) nonetheless LLMs are most important technology since the Internet 3) however this tech isn’t safe many people will get shot by the algo

random hot take: the rebuild Silicon Valley in Miami nutters would have had far more traction if they had gone for Pittsburgh instead. the CMU talent pool is crazy deep. I can't believe I'm saying that bc, well, Pittsburgh. but I think it would have made much more sense.

Grok is like your dumbest friend from high school

this is not going to age well

this is not that bad actually bc of the lack of depth, I can probably do this or very close to it without even needing to use belt sleeves wraps anything. it’s much easier than it looks.

the best part about this tweet is that she correctly called it a strike

gm

yeah OpenAI nerfed the earlier models out of concern for their users psychological well-being, that sounds so on brand for them 🤣