开启时会模糊预览图，关闭后正常显示
居然还有人做搞抽象的插件……
Believing is seeing
RT @cornu__copia: @bronzeagemantis A lot of the Anglosphere urban dysfunction boils down to the fact that they are both numerically illiter…
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
有个朋友休息了一年，准备找工作，找了一段时间还没有合适的 😂产品经理年薪六十万的工作，感觉就挺难找的，去年我和她说，现在这个公司的待遇已经非常好了，出去很难找到这样待遇的工作了。高收入久了之后，对市场的感知也没那么敏锐了，想找合适的创业公司，只是哪家创业公司愿意出这个价呢
程序员，开发过 2 个 iOS App，几个不知名的浏览器插件，几个小程序。喜欢看动画、漫画、轻小说和网文。希望早日成为自宅警備員。
Sam at least didn't pivot on Trump in 3 hours
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
!["Out model blackmailed our researcher over their affair"
"Our model tried to get out of the docker container"
😂 - it looks like only models in America do all these things, whereas models from China work like diligent knowledge factory workers!!!!
Should we ban AI research in the western hemisphere to protect humanity?
At this point, I have come to sincerely believe AI Security Research is an elaborate scam.
I am not talking about interpretability research, which is awesome & important to understand internal behavior and to reduce hallucinations.
AI security research, on the other hand, is an art to write stories that will sound convincing to boomer politicians & decision makers & common people based on their perception of how AI systems work.
Common people believe, for example
- models have infinite context, like humans
- their interactions with different users result in single unified context
- in fact, they believe there is a single instance of the model (like skynet) that is answering all the users at once. [on this one, please include Paul Graham as well]
Only if those things were true, the below news would make sense. But it is not.
But, beautiful fiction writing. I will give you that.](/_next/image?url=https%3A%2F%2Fpbs.twimg.com%2Fprofile_images%2F1882081691973586944%2F6Eqmsf_y_400x400.jpg&w=3840&q=75 "\"Out model blackmailed our researcher over their affair\"
\"Our model tried to get out of the docker container\"
😂 - it looks like only models in America do all these things, whereas models from China work like diligent knowledge factory workers!!!!
Should we ban AI research in the western hemisphere to protect humanity?
At this point, I have come to sincerely believe AI Security Research is an elaborate scam.
I am not talking about interpretability research, which is awesome & important to understand internal behavior and to reduce hallucinations.
AI security research, on the other hand, is an art to write stories that will sound convincing to boomer politicians & decision makers & common people based on their perception of how AI systems work.
Common people believe, for example
- models have infinite context, like humans
- their interactions with different users result in single unified context
- in fact, they believe there is a single instance of the model (like skynet) that is answering all the users at once. [on this one, please include Paul Graham as well]
Only if those things were true, the below news would make sense. But it is not.
But, beautiful fiction writing. I will give you that.")
"Out model blackmailed our researcher over their affair" "Our model tried to get out of the docker container" 😂 - it looks like only models in America do all these things, whereas models from China work like diligent knowledge factory workers!!!! Should we ban AI research in the western hemisphere to protect humanity? At this point, I have come to sincerely believe AI Security Research is an elaborate scam. I am not talking about interpretability research, which is awesome & important to understand internal behavior and to reduce hallucinations. AI security research, on the other hand, is an art to write stories that will sound convincing to boomer politicians & decision makers & common people based on their perception of how AI systems work. Common people believe, for example - models have infinite context, like humans - their interactions with different users result in single unified context - in fact, they believe there is a single instance of the model (like skynet) that is answering all the users at once. [on this one, please include Paul Graham as well] Only if those things were true, the below news would make sense. But it is not. But, beautiful fiction writing. I will give you that.
AI @amazon. All views personal!
I'd vote for anyone for mayor of New York over Mamdani. Even a robot. https://t.co/rtUzuIf1HF
Professor of computer science at UW and author of '2040' and 'The Master Algorithm'. Into machine learning, AI, and anything that makes me curious.