SemiAnalysis reports that US smartphone imports from China have dropped from 90 percent to 25 percent, eroding the centrality of China's Foxconn assembly network for Apple. The thread frames it as a forced derisking of the consumer electronics supply chain.
US smartphone imports from China have collapsed from 90% to 25%.
The Foxconn China assembly network, once the undisputed backbone of Apple's hardware empire, is seeing its centrality eroded in real time. This is the clearest data point yet of a forced, systematic derisking of the consumer electronics supply chain. (1/3)🧵
Vishal Misra argues that the 1948 Bell Labs paper by Claude Shannon, A Mathematical Theory of Communication, contains the first generative AI model and that modern LLMs trace back to it. He quotes a thread by techNmak recounting Shannon's history.
1/2 This 1948 paper also has the first “generative AI” model of any kind - all modern LLMs can trace back their origins to what was written in this paper.
Nav Toor describes WeClone, a self-hosted AGPL-3.0 project with about 16,400 GitHub stars that exports chat logs, fine-tunes an LLM on a user's messages and binds it to a chatbot. The post presents it as a free alternative to commercial digital persona services.
Someone built a tool that reads all your chat messages and creates an AI clone of you. It talks like you. Responds like you. Thinks like you. 16,400 GitHub stars.
It's called WeClone.
Export your chat history. Feed it to the tool. It fine-tunes an AI model on YOUR messages. Your slang. Your humor. Your tone. Your personality. Then it binds to a chatbot and becomes you.
Not a generic chatbot with your name on it. An AI trained on thousands of YOUR actual conversations. It learns how YOU respond to questions, jokes, arguments, and small talk.
Here's how it works:
→ Export your chat logs from WeChat, Telegram, or any messaging app → WeClone processes and cleans the data automatically → Fine-tunes an LLM on your conversation style and patterns → Captures your unique vocabulary, tone, humor, and personality → Binds the trained model to a chatbot interface → Your digital twin is live. People can talk to "you" when you're not there.
Here's the wildest part:
Your friends text your AI clone. They can't tell it's not you.
It uses your actual phrases. It mirrors your response timing patterns. It knows how you react to specific topics because it learned from real conversations where you did exactly that.
This is not a parlor trick. This is digital identity preservation. Your grandchildren could talk to an AI version of you long after you're gone. Your personality. Your stories. Your humor. Preserved.
The AI twin industry is projected to be worth billions. Companies charge thousands for custom digital personas.
This is free. Self-hosted. Your data stays on your machine.
Quant investor qm, who says he is retired from Renaissance Technologies, shares an unseen interview snippet from a former colleague named Nick and promises valuable insights. He also introduces a planned series on quant practitioners and resources, linking an image of a recommended website.
some have asked me about my time in Renaissance Technologies. although I’m retired I can’t really say much due to NDA, but I have an unseen interview snippet from my ex-colleague Nick (hope the kids are doing well mate) that I’m comfortable to share. a lot of alpha in there
I find the quality of content on QuantTwitter disappointing. so in the next few weeks, I will be sharing novel stories about practitioners/legends, resources, anecdotes & many more
to kickstart this initiative I would like to share one of the most valuable websites for quant:
