Karpathy分享语音“长篇倾诉”技巧：让LLM整理混乱思路

## Combined views 3.3M 14 posts, first seen 11d ago 3.3M 14 posts, first seen 11d ago ## Sentiment ### Positive Read Many users praise voice rambling sessions with LLMs as game-changing and more accurate than typing for better prompting, while others dismiss them as gimmicks due to latency and transcription flaws. Based on 544 visible X reactions from 2340 accounts. ## Useful Links karpathy.ai Silicon Republic One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10… Show more ## Sentiment ### Positive Read [...] # Eureka Labs' Andrej Karpathy prompts LLMs with 10-minute voice rambles LLMs successfully organize the chaotic, stream-of-consciousness input. One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs mo

This approach works very effectively. It also helps you be much more prepared when using your lawyer’s time to finalize such documents. ##### 11. Start a new chat Conversations that get too long get unwieldy for the LLM. Any previous wrong turns continue polluting the context and confusing the LLM. If you find yourself saying “try again”, it’s better to start with a new conversation. Ask the LLM to summarize your previous instructions in the old chat, and copy those over as a jump-off point for the new conversation, so you’re not starting from scratch. Sometimes, the LLM just needs to traverse a new path down the probability tree of possible tokens. ##### 12. Start many chats [...] ##### 5. Define what NOT to do Left unchecked, LLMs will make a best guess. And their training makes them lean towards pleasing you. Draw constraints: “Push back if I’m wrong”, “Don’t be deferential”, “Don’t jump to solutions, but give me options”, “Don’t avoid or explain technical jargon”, “Don’t use f

| | | | --- | | | cruffle\_duffle 9 months ago | root | parent | next (javascript:void(0)) That rubber duck is a valid workflow. Keep iterating at how you want to explain something until the LLM can echo back (and expand upon) whatever the hell you are trying to get out of your head. Sometimes I’ll do five or six edits to a single prompt to get the LLM to echo back something that sounds right. That refinement really helps clarify my thinking. …it’s also dangerous if you aren’t careful because you are basically trying to get the model to agree with you and go along with whatever you are saying. Gotta be careful to not let the model jerk you off too hard! | [...] | | | | | --- | | | cruffle\_duffle 9 months ago | root | parent | next (javascript:void(0)) That rubber duck is a valid workflow. Keep iterating at how you want to explain something until the LLM can echo back (and expand upon) whatever the hell you are trying to get out of your head. Sometimes I’ll do five

“One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.” [...] "One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. I

The AI Advantage LeaderboardsAbout AI Advantage Team AI Advantage Team 7d • AI News 📰 AI News: Karpathy's Ramble Technique: Talk Messy for Ten Minutes, Let the AI Clean It Up 📰 📝 TL;DR 📝 Andrej Karpathy shared a simple prompting habit worth trying today: switch to voice input, ramble for about ten minutes about a messy problem with zero filter, and let the AI reconstruct your scattered thoughts into something clearer than what you started with. No new tool, no product launch, just a genuinely useful pattern anyone can test right now with voice input already built into ChatGPT, Claude, or Gemini. 🧠 Overview 🧠 This is commentary, not a launch, and it's worth being upfro... See more 2 1 comment AI Advantage Team

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.