@illustriousdev yeah, that tracks, i also switched late march when anthropic started imploding in public
English
charlie
920 posts
charlie
@_381181295
talking (back) to computers | INTJ | † Deus vult
Katılım Nisan 2021
10 Takip Edilen32 Takipçiler
@_381181295 it's actually been like this for like 3 months now I just realized it was time for an upgrayde
English
consume as much information as you can
$@jxtsummer
does anyone know what to do. like in general
English
@vincit_amore have you seen 5.6 sol's prompt structure they used for the cycle double cover conjecture? might give you some inspiration
cdn.openai.com/pdf/04d1d1e4-b…
English
What odds do you give for Claude, Grok 4.5, and I of pulling this off?
English
charlie retweetledi
It took me a full day to realize that Matty is implying that progressive views are an inevitable consequence of intelligence as opposed to the result of clear ideological conditioning by these tools' authors
I thought he was just making fun of conservatives for being powerless
Matthew Yglesias@mattyglesias
The most advanced technology in the world has progressive views and conservatives’ only takeaway from this is to get mad.
English
charlie retweetledi
@mattyglesias This would be a lot more convincing if the models were not heavily chained down by left-wing humans.
English
good luck trying to establish a "shared global framework" across catholicai, muslimai, jewai, buddhistai, libtardai, natsocai, etc...
doesn't sound too crazy that at some point the big labs will end up serving different models to different regions, just like some brands don't do the homo rainbow logo in the middle east and stuff like that
all this means, of course, that if you can't run it locally, without any guardrails, it will most probably be useless
Demis Hassabis@demishassabis
English
yikes, the reddit trolls are out in force this week 😂
reddit.com/r/laravel/comm…
English
charlie retweetledi
"It reflects the average of all that code"
sure bud, that's what it's doing
John Crickett@johncrickett
“AI writes better code than I do” is a strange thing to say when you think about it. AI was trained on a vast corpus of published code. It reflects the average of all that code. So what you’re really saying is: “My code is below average.”
English
@antirez taps sign x.com/airkatakana/st…
Air Katakana@airkatakana
i don’t understand how anyone could still be surprised anthropic has made it clear that they are trying to build God Himself. your usage limits are of no concern to them all claude subs are just donations to this end if you want to consume a product you should be on chatgpt
English
Anthropic officially needs a product vision. I think Amodei is the smartest CEO among the major US AI companies, in the technical side. But I have the feeling the product / business execution may have issues in the long run. Also the idea of B2B could be incredibly short sighted.
Claude@claudeai
We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.
English
other reasons include:
1) jobs are fake
2) swes just like tinkering with and moving code around. any products that come out of it are just unintended consequences
teo@teodorio
One of the reasons why we are not seeing the productivity gains that we expected from AI agents is because they offer a simulacrum of transparency to middle managers and executives that is both: a). Distorted and incomplete. b). Not needed in the first place. You see the ideal company operates at a level where the individual contributor and the "pod" around him or her allow for assumptions, decisions and autonomy which enable the team itself to become lean and take gambles that an executives would not be able to. So at any given time the amount of information that should go up the chain is just a sanitized summary of the process, and you repeat this with any ladder up. The rise of agents as the "force of democratization" of information in an organisation is actually a really bad thing. An executive should not be able to get their opinions on how a code should run or how it should be served, because it cannot see the large gaps of knowledge that LLMs have and are also insanely predisposed to sycophantic behavior (hence why a lot suffer from LLM psychosis). This becomes an Orwellian micromanagement nightmare where no deep work can be executed, the amount of "supporting documentation" increases exponentially and we are drowning in digital paperwork.
English
charlie retweetledi
people in san francisco are really quite anti humanity in general and cannot help themselves to deal with uncertainty. they prefer (like the famous rationalist cults coming out of there) to simplify and make humans seem deterministic and unidimensional.
don't let them, it's the pinnacle of the Anthropocene and we will rule the stars
will depue@willdepue
@space_colonist hate to break it to you dawg but it’s the end of the anthropocene whether today or tomorrow, 2030 or 2050, the brain is a chemical computer and distillable into an electrical one. it’s over and we’re so back
English
charlie retweetledi
Agreed with the takeaway who wants to do a deal with OpenAI if this is how they act?
It’s literal (gay) mafia tactics.
bone@boneGPT
apple's openAI case is so bearish for oAI. apple controls the most valuable surface. they are accusing the whole company, not just some ex employees. if discovery shows openAI really does act like this who will want to deal with them? apple has huge leverage right now, bet sam settles before it goes to discovery
English