I’m trying to decide which audience has gotten more drunk off of its Kool Aid — the crypto people who took “decentralization” too seriously or the AI people who have taken the risks of a super intelligent AGI too seriously.
I suspect the AI folks have it worse as crypto is useless as a technology so it’s been easy to write them off but since LLM’s are actually useful we have to humor the AI crowd’s delusions of grandeur. So they sink deeper into the delusional thinking and ignore real problems
I’m reminded of the naivety of early social media engineers assuming that if you connect the whole world then we’d live happily ever after. Instead we now know it’s just a thankless job fighting hate speech, election meddling, COVID misinformation, cyber bullying, etc.
The AI people are actively ignoring thinking through the harms LLMs can and are causing today to focus on fantasies inspired by books and movies they saw as kids.
@carnage4life I agree with you. Crypto madness was somewhat self-contained because it depended on creating a fog of uncertainty around things like value and currency. But we actually do understand those things pretty well, so there was limited space for the crypto game to be played.
We do not, in contrast, understand things like intelligence, sentience and consciousness very well. There's a lot of room for uncertainty, self-delusion, well-intended error and toxic bullshit to romp about in.
@carnage4life a fantastic post about this very topic, only 9000 words, but worth it:
@carnage4life Let's not forget that there is a fairly large overlap between those groups.
Can't wait for AI-crypto kids. "Let's use AI to make VR NFTs. Going to the moon!"
@carnage4life Thank you for acknowledging that the tech IS still useful despite how many of the folks building it have bought into their own hype.
This tech makes my life, and work as a knowledge worker, better every day; it’s SO vastly weird watching other folks claim it’s useless because it ’just predicts the next word’.
There ARE people building interesting stuff on LLMs and diffusion models who aren’t bought into AGI and just want to make better, useful, interesting, and beautiful tools. 🤷
Dare Obasanjo
Ryan Marx
Philippe Paquet
Richard Grant
Howard Chu @ Symas
Ivan Andriollo 
MaybeMyMonkeys
Christo
She smashes
Dominic Hopton
Matt May
Cypherfox