Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)L
Posts
2
Comments
104
Joined
2 mo. ago

  • also this post which is where I got the xAI co-founder statement from, also goes over other things

    -the Anthropic team lead quitting (which we already discussed in this thread)

    -AI is apparently so good a filmmaker with 7 years of experience said it could do 90% of his work (Edit: I thought this model was unreleased, it’s not, this article covers it)

    -The Anthropic safety team + Yoshua Bengio talking about AIs being aware of when they’re being tested and adjusting their behaviour (+ other safety stuff like deepfakes, cybercrime and other malicious misuses)

    -the US government being ignorant about safety concerns and refusing to fund the AI international report (incredibly par for the course for this trash fire of an administration, they’ve defunded plenty of other safety projects as well)

  • another co-founder has quit praises Elongated Muskrat (lmfao) and says recursive-self improvement in the next 12 months and 100x productivity real soon (alongside those self-driving cars Musk promised back in 2012)

  • read the whole article, of course Donald Trump is excited about this. Which also guarantees that this will crash and burn in no time at all

  • well that headline just filled me with dread

  • neither did I reading this article was my first exposure to him

  • the aforementioned wikipedia page has got some criticisms of his works under the critical reception section

  • this article involving an incredibly eyebrow-raising take from one of the people at METR (the team behind the famous "tasks AI can do doubles every 7 months" graph) saying AI is eventually going to become more impactful than the invention of agriculture and more transformative than the emergence of the human species and also calls it an intelligent alien species. Immensely funny amongst the other people saying "please stop treating AI like magic"

    the Harari guy also seems to be into transhumanism if a skim of his wikipedia page is correct. The “this is the first time in history that we have no idea what the world will look like in 10 years” thing is also an eyebrow-raiser. I could probably rattle off a couple examples (ie the two world wars)

  • OpenAI is probably toast tldr OpenAI's financial situation is more cooked as a big investor shows doubt, WeWork 2 imminent

  • Deploying chatbots into high-stakes situations with absolutely zero safety regulations. Surely this will have zero consequences and go completely fine

  • "we're scared we've crossed a line"

    ...seeking money from the UAE and Qatar didn't count as crossing a line?

  • just one more data centre's gonna do it! just give me a couple million more bucks!

  • also completely leaving out important context on the Iran/stuxnet example, in that it was a joint effort between two countries believed to have been in development for five years. The idea that AIs will engage in lightspeed wars and disable all critical infrastructure in a single day while speaking in alien languages and creating alliances is unreasonable extrapolation of the capabilities. Also completely ignored the segment where the Anthropic team implemented safeguards and communicated with the teams behind the software to patch out the bugs. It's the most blatant fearmongering ever. Thank god the comments contain reasonable responses and breakdowns of the post. That channel's way of highlighting papers just pisses me off

  • community posts have been a thing for like, two years now? three?

  • Eliezer’s latest book says humans and chatbots are both “sentence-producing machines” so yes

  • Just decide to be sane, sanity is a skill issue

    Yud has officially cured mental health forever, psychologists and therapists in shambles

  • 2034 is also the year superintelligence is gonna happen according to the updated predictions from the AI 2027 crew, so double whammy!

  • Fuck, I was just about to post that. You beat me to the sneer

    chatbots doing normal chatbot things