China just passed a law to mobilize its entire society for war — industry, tech, healthcare, transport, energy, civilians. All of it.
This is not a drill. This is legislation. 🧵
#China#Taiwan
On August 28, China’s NPC Standing Committee approved a comprehensive revision of the National Defense Mobilization Law (中华人民共和国国防动员法). It takes effect October 1.
#davidmelaracom#ChinaTaiwan
@gdb Claude Code on desktop now has a built-in browser. Imagine ChatGPT seeing the same webpage as the user in real time, understanding the current context, and acting as a mentor while you navigate—no screenshots, no copy-paste, no switching tabs. Please bring this to ChatGPT & Codex
@gdb@sama@OpenAI@gdb@kevinweil
Claude Code on desktop now has a built-in browser. I’d love ChatGPT to go one step further: see the same webpage as the user, understand the live context, and guide them while they navigate. No screenshots. Just shared context.
@gdb AI shouldn’t need screenshots to help.
The next generation of AI assistants should share the user’s live context.
Imagine ChatGPT seeing the same webpage you’re seeing, understanding where you are, and guiding you step by step as you navigate—without screenshots or copy-paste.
AI shouldn’t need screenshots to help.
The next generation of AI assistants should share the user’s live context.
Imagine ChatGPT seeing the same webpage you’re seeing, understanding where you are, and guiding you step by step as you navigate—without screenshots or copy-paste.
@sama@OpenAI@gdb@kevinweil AI shouldn’t need screenshots to help.
The ideal assistant should see the same webpage you’re looking at, understand the live context, know where you are, and guide you as you navigate—like a mentor sitting next to you.
Shared context is the next UX breakthrough for AI.
@sama@sama@OpenAI@gdb@kevinweil
Claude Code on desktop now has a built-in browser. Imagine ChatGPT seeing the same webpage as the user in real time, understanding the current context, and guiding users as they navigate—no screenshots, no copy-paste, just shared context.
@sama Hey @sama, @OpenAI, @gdb, @kevinweil,
Claude Code on desktop now has a built-in browser. Please bring the same capability to Codex: browse, click, scroll, inspect pages, fill forms, and debug apps—all without leaving the workspace.
@sama@huggingface Hey @sama, @OpenAI, @gdb, @kevinweil,
Claude Code on desktop now has a built-in browser. Please bring the same capability to Codex: browse, click, scroll, inspect pages, fill forms, and debug apps—all without leaving the workspace.
@sama@gdb@kevinweil@OpenAI@sama@gdb@kevinweil@OpenAI That's exactly the issue. The model understood the request. it was stopped by response-generation limits. ChatGPT interrupts the workflow and forces users into multiple interactions for tasks that should be completed in one continuous flow.
@sama@gdb@kevinweil@OpenAI@sama@gdb@kevinweil@OpenAI I asked it to transcribe my sermon, ChatGPT replied:
"The complete sermon is too long to reproduce in a single response. The manuscript exceeds the maximum response length I can generate. The only way is to split it into multiple parts..."