LogoThread Easy
  • 探索
  • 撰写 Thread
LogoThread Easy

您的一体化 Twitter 线程助手

© 2025 Thread Easy All Rights Reserved.

探索

最新在前,按卡片方式浏览线程

开启时会模糊预览图,关闭后正常显示

I log my sessions with /end-session to log:

- overview of what we did
- files we created/deleted/modified
- why we made decisions
- what worked/didn’t work 
- steps for next session

then I run /start-session with new agent and it reads the last log to get the context it needs

I log my sessions with /end-session to log: - overview of what we did - files we created/deleted/modified - why we made decisions - what worked/didn’t work - steps for next session then I run /start-session with new agent and it reads the last log to get the context it needs

TLDR; I make software with AI and talk about it. Serial internet biz builder with multiple 6 & 7 figure exits. Always learning.

avatar for Ian Nuttall
Ian Nuttall
Mon Dec 22 09:09:20
100% 客户端 JS 运行
Gemini Watermark Remover 一个基于纯 JavaScript 开发专门用于移除 Google Gemini AI 生成图片右下角的水印。

无损直出:没有 AI 幻觉,没有模糊,像素级还原,就像水印从来没存在过一样。

https://t.co/Kat9le6FWx

100% 客户端 JS 运行 Gemini Watermark Remover 一个基于纯 JavaScript 开发专门用于移除 Google Gemini AI 生成图片右下角的水印。 无损直出:没有 AI 幻觉,没有模糊,像素级还原,就像水印从来没存在过一样。 https://t.co/Kat9le6FWx

🧠在家居士 | 🥦素食者 | 🏃🏻马拉松爱好者 | 💰省钱小能手 | 搭🪜技术资深学者 | 👨‍💻科技宅 | 🆕更新狂 | 🆅 六边型战五渣

avatar for Geek
Geek
Mon Dec 22 09:07:23
Factory should let you publish the chat threads like Amp does

would be cool to see the prompts to build a design like this (because I hate designing stuff)

Factory should let you publish the chat threads like Amp does would be cool to see the prompts to build a design like this (because I hate designing stuff)

TLDR; I make software with AI and talk about it. Serial internet biz builder with multiple 6 & 7 figure exits. Always learning.

avatar for Ian Nuttall
Ian Nuttall
Mon Dec 22 09:02:44
I've been an LLM pessimist for years but I think now that LLM and reasoning tools have now reached a point where they may genuinely transform science and engineering. I do not believe this is AGI, but rather something like bringing generalizing past tools like Google Search or SAT solvers to broader domains. 

This raises some questions. In coding and writing, the practice of the discipline looks to be transforming entirely. Policing students from not using AI seems difficult, but using AI could genuinely undercut beginner understanding and growth. At the same time, AI looks to be a powerful force multiplier for the expert. I have been asked how students should navigate this situation. My two cents is learning to write and code without AI is probably crucial for the best results.

Add onto this other risks of AI such as LLM psychosis. We will see many more beginners claim "breakthroughs" in disciplines they don't understand. This is a shame since it will probably trigger a fallback to credentialism. Unfortunately there are not enough hours in the day for experts to scrutinize beginner work. Until LLM world models grow sophisticated enough to self verify, this is probably unavoidable. 

What does this mean for science or engineering? I am perhaps a pessimist but I think the singularity is nowhere to be seen. Rather I think AI inefficiency will nearly counterbalance AI gains leading to modest gains. Speaking for myself, I have been moving slower on @deep_chem PRs since I have more distrust of contributed code. Students try to use AI and contribute incorrect code. As a result, I only merge in vetted code by proven existing contributors... AI has slowed progress by raising code review burden. This is an anecdote but I see similar patterns playing out elsewhere. At the same time, experts will use AI to achieve breakthroughs. Increasing efficiency and inefficiency will lie in a tenuous balance.

The problem of developing automated verifiers consequently becomes the biggest challenge in AI. I suspect this will be a long slow grind, but there lies the path to real breakthroughs. 

(As a last comment, this post is entirely AI free! Avoiding AI leads to crisper thought when brainstorming)

I've been an LLM pessimist for years but I think now that LLM and reasoning tools have now reached a point where they may genuinely transform science and engineering. I do not believe this is AGI, but rather something like bringing generalizing past tools like Google Search or SAT solvers to broader domains. This raises some questions. In coding and writing, the practice of the discipline looks to be transforming entirely. Policing students from not using AI seems difficult, but using AI could genuinely undercut beginner understanding and growth. At the same time, AI looks to be a powerful force multiplier for the expert. I have been asked how students should navigate this situation. My two cents is learning to write and code without AI is probably crucial for the best results. Add onto this other risks of AI such as LLM psychosis. We will see many more beginners claim "breakthroughs" in disciplines they don't understand. This is a shame since it will probably trigger a fallback to credentialism. Unfortunately there are not enough hours in the day for experts to scrutinize beginner work. Until LLM world models grow sophisticated enough to self verify, this is probably unavoidable. What does this mean for science or engineering? I am perhaps a pessimist but I think the singularity is nowhere to be seen. Rather I think AI inefficiency will nearly counterbalance AI gains leading to modest gains. Speaking for myself, I have been moving slower on @deep_chem PRs since I have more distrust of contributed code. Students try to use AI and contribute incorrect code. As a result, I only merge in vetted code by proven existing contributors... AI has slowed progress by raising code review burden. This is an anecdote but I see similar patterns playing out elsewhere. At the same time, experts will use AI to achieve breakthroughs. Increasing efficiency and inefficiency will lie in a tenuous balance. The problem of developing automated verifiers consequently becomes the biggest challenge in AI. I suspect this will be a long slow grind, but there lies the path to real breakthroughs. (As a last comment, this post is entirely AI free! Avoiding AI leads to crisper thought when brainstorming)

Founder and CEO @deepforestsci. Creator of @deep_chem. Author @OReillyMedia. @stanford CS PhD. https://t.co/7LDcegrCsc

avatar for Bharath Ramsundar
Bharath Ramsundar
Mon Dec 22 09:01:56
there's only one thing you should adopt from this Agile Transformation and it's:

"Every day. We stand. For 45 minutes."

stand more, sit less, ditch the rest, and focus on what matters

there's only one thing you should adopt from this Agile Transformation and it's: "Every day. We stand. For 45 minutes." stand more, sit less, ditch the rest, and focus on what matters

building @VemetricHQ & @snappify_io

avatar for Dominik Sumer ✨
Dominik Sumer ✨
Mon Dec 22 08:54:04
RT @DeepakNesss: Basically, one web scraping company is suing another scraping company for scraping the scraped data.

RT @DeepakNesss: Basically, one web scraping company is suing another scraping company for scraping the scraped data.

Running 10+ SaaS in parallel at https://t.co/3sUSmGdehM since 2011! ⚡️ 🇮🇹 Italian 🎷 Jazz musician 🥎 Padel addicted

avatar for Mike Rubini
Mike Rubini
Mon Dec 22 08:34:56
  • Previous
  • 1
  • More pages
  • 153
  • 154
  • 155
  • More pages
  • 5634
  • Next