infrecursion

[infrecursion @infrecursion1]

@jachiam0 There is so little evidence of Claude models generalizing well outside their training domain and recent models seem to be more and more benchmaxxed. They certainly rely a lot on "mystery" and "marketing" to hype their models.

[infrecursion @infrecursion1]

@jachiam0 I don't know about "vibes for trust and governance," but just for "vibes of raw intelligence," Anthropic is nowhere near the mandate of heaven. Their models are only good at inducing deep psychosis and dependence in most users. I have yet to see Claude solve an open problem.

[infrecursion @infrecursion1]

@JohnHolbein1 There should be two more options that say "I don't trust Pangram at all to ascertain AI writing" and, most importantly, "I read and recommend a paper based on their quality, not based on who wrote it, so I couldn't care less about Pangram score".

[infrecursion @infrecursion1]

@jenzhuscott Why tf would I use someone else's vibe coded app when Claude/codex can easily make me whatever I want or need anytime?

[infrecursion @infrecursion1]

@tracewoodgrains Same to you lol. Stop pretending you wrote something deep is likr that meme I am 14 and this is deep.

[infrecursion @infrecursion1]

@nxthompson Here it feels like the biggest leap in AI use case is spamming every post with AI replies.

[infrecursion @infrecursion1]

@burny_tech No one teaches anyone "niche phd topics". You just read papers and do research until you get it or give up.

[infrecursion @infrecursion1]

@ohryansbelt @paulg They should do whatever the fuck they want to instead of listening to gatekeepers.

[infrecursion @infrecursion1]

@ohryansbelt @paulg Lol dangerous corner for whom? If we accept that superhuman ASI will be there in like 5 years, what difference will it make what they do? There will not be a single damn thing they can do even remotely better.

[infrecursion @infrecursion1]

@mwstateofmind @___frye Lmao you're a fucking moron. Do you even know what stealing means or how LLMs are trained?

[infrecursion @infrecursion1]

@eliebakouch The dreaming system has been in ChatGPT for about a year now.

[infrecursion @infrecursion1]

@renegadesilicon @Marshwiggle119 Your entire post is a handwave. So basically you're saying counter my bullshit with more bullshit. Just stop and accept you don't know shit about any of this.

[infrecursion @infrecursion1]

@yeehahboogie @gilbert I think some of the registers cycle by default? Anyway, Emacs with kill-ring (with similar plugins in Vim) is also there, so there is fear of storing anything in brain.

[infrecursion @infrecursion1]

@majavonwestphal These cars are impossible to drive. There is no place to put harnesses. How will the horse even drag this thing? Driving a car is like trying to operate on a carriage whose wheels, seats, and roof are all compromised. There is nothing to leave intact, nowhere to begin.

[infrecursion @infrecursion1]

@TSoS_ Lmao, the fucking arrogance about cognitive decline from someone who has achieved literally zero significant thing in their life and never will. You were nothing with or without AI. Go fuck yourself.

[infrecursion @infrecursion1]

@thsottiaux I think the name Codex is older than ChatGPT for these kinds of things, so it also should have its place in history.

[infrecursion @infrecursion1]

@NitroSpidey Yes everyone who has actually achieved something seems to have been brainwashed. Not you though, your complete lack of talent and intelligence helps shield you from genai and other bs.

[infrecursion @infrecursion1]

@PhilipDBunn More like covid showed how fragile the school system is before ChatGPT showed it was obsolete.