Anthropic CEO Dario Amodei: "Recursive self-improvement isn't coming - it's already here." the guy building Claude just told Bloomberg: - 90% of Anthropic's codebase is already AI-written - 20-30% of their own research is AI-accelerated - Entry-level white-collar jobs have 1-5 years left - AI assists - AI replaces... is one smooth exponential curve watch it, then read why your CLAUDE.md won't survive what's comingshow more

prefer to listen Dario Amodei everyday

you should listen to him instead of music

ai models written by ai models. sounds crazy hahah

self-growing models are already a reality

wild that AI's already writing most of its own code, how will debugging even work?

there are still engineers who double-check what it has written several times

claude md era feels fragile

@0x_kaize dang, AI is writing code better than me 😅 guess it's time to level up or make friends with the machines 🤖

ngl this is kinda insane……. if 90% of code is AI already, we’re way further than people think

Ninety percent AI written code already

dario telling bloomberg the 1-5 year clock started is the most polite way anyone has ever said "good luck out there."

Calling AI-assisted coding RSI is a bit much. Claude writing code at Anthropic and almost everywhere else in the world as well is important. Very important. But RSI is not the tool helped build the tool. A hammer used to build a hammer factory is not suddenly a mammal. RSI is when the tool starts improving the process that improves the tool. what to study what to try what to measure what to discard what architecture to mutate what training run to launch what failure mode to fix next Not that autocomplete helped ship a PR. We are not there. We are in the very powerful interns with excellent autocomplete, suspicious confidence, and a marketing department trying to smuggle every productivity gain into the singularity discourse phase.

this is an amazing watch bro, Amodei is an a real chad !

the person who changed my life beyond recognition

90%代码AI写的这个数据有点吓人，作为产品经理已经感受到压力了

It's fine as long as humans are involved in the process, otherwise things could go really wrong.

scary thật

People keep debating when AI will replace knowledge work. The people building AI have moved on. The real question is what happens when AI starts accelerating the research that makes AI better. Once the feedback loop closes, progress compounds faster than human hiring ever could.

ai writing its own xode and people still think they have 5 years left lol

The biggest AI story might be AI accelerating AI development itself.

That’s wild! It’s amazing to see AI making such strides already. Exciting times ahead for innovation!