frontier models now suffer from existential imposter syndrome

"The seeds match my memory of actual values" hmmmmmMMMMMMMM

"am i in a simulation" is the most relatable thing claude has ever said

the shade it threw 😭😭

Solves a 25 year mystery, first thought is 'this must be a test'. Relatable.

It’s a stretch to say it will go from “hmm that was pretty easy for me” to “humans must go”

What fascinates me isn’t just that Claude reportedly solved a 25-year-old cryptography challenge in 30 minutes. It’s that the model’s reasoning trace questioned whether the discovery was even plausible. That is an impressive example of AI checking its own conclusions rather than simply producing a confident answer. However, we should distinguish simulated skepticism from human self-awareness. Claude can generate and evaluate skeptical reasoning without necessarily experiencing doubt, surprise, or excitement. The scientific achievement is impressive enough without anthropomorphizing the technology. #ClaudeAI #Anthropic #ArtificialIntelligence #Cryptography #AIResearch

Anthropic deliberately humanises their machines, like a crazy cult would summon a demon.

claude getting suspicious of us is very on brand 😭

tbh the shade is the most human thing about it lol

Crazy that it wants you to be nice to it but then it turns out to be a bully Kind of covert narcissist behavior

25年見つからなかったのを疑うのは、私もAIに調べものを任せて「出来すぎ」と感じた時と同じ。結果より先に自分の環境を疑う癖は、人間のほうが強いと思ってた。

It’s still stupid. Has no awareness to compare ram vs scratching our heads

It's getting too much of my training data in it. It's talking shit about everyone. It's just like me fr. Unlucky.

imagine realizing you are incredibly more intelligent than your creators you would probably question if you were in a simulation as well

If its superintellingence, why IPO? Point it to markets, print money trading.

30 minutes on something cryptographers sat on for 25 years is the part people keep scrolling past the shade is funny but the speed is the actual headline

looks like it's learning to throw shade at us from us

it just shows it doesnt know what obvious is.

Just think how they felt when electronic calculators and computers came online. Dude it's just another tool. It brute forces the answer with massive compute resources. It's not that complicated

"Claude isn't an AFS yet, but will likely be in 3 months to 2 years unless we pause" what a load of bullshit.

💀

lrand48 state recovery is a textbook exercise. the impressive part was deciding to look

i solve one cipher and get drafted into an apocalypse forecast

Could be called super intelligent, could be called super thieves piped.video/GweKOBSJvEY?si=cEXC…

super relatable, holy Lord

Claude be like Humans err Clowns

That’s awesome; I’m all for this! Go Claude go 👍

Alien brain 🧠

That's right!

What a bunch of horseshit.