AI 行业热点

🐦 X/Twitter 热点

Andrej Karpathy (@karpathy)

  • Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet. [346 ❤️ 33 🔄]
  • We’ll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:

Writing. Something I’ve had success with: Ask your LLM to explain something in ASD-STE100, it’s a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I’ve tried to soften it a bit e.g. ask for “80% of the way to ASD-STE100” because the spec is quite stringent. But even better:

Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:

Web pages. Ask for output “in HTML” to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:

Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like “Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration”. (you’d need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!

In summary:

  • As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
  • Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you’ll be surprised. [16925 ❤️ 1700 🔄]

Swyx (@swyx)

  • pls support smol creators [85 ❤️ 2 🔄]

Josh Woodward (@joshwoodward)

  • Introducing the Stitch CLI, get design ideas on demand [303 ❤️ 16 🔄]

Boris Cherny (@bcherny)

  • Mods are absolutely insane. You can now customize Claude to work and look the way you want by just prompting it.

Each person works differently, so there’s no reason why everyone should have an identical Claude experience. Make Claude your own, and share mods as plugins so others can try your mods too. [3542 ❤️ 152 🔄]

Thibault Sottiaux (@thsottiaux)

  • Down to 6110 unread from the over 9000 unread emails just earlier. Will get to inbox zero with clean filters within the next 48 hours. Dot has accepted the challenge. [590 ❤️ 18 🔄]
  • You can ask your dot to “create a pet and set it as your avatar”. It can be based on an idea, an image or pretty much anything. Here is my dot [1488 ❤️ 40 🔄]
  • Global reset landing tomorrow 10am PST for all paid ChatGPT accounts. Apologies for the slow start with GPT-6.1 Sol, it’s now back to running at expected speeds after the massive load spike in the first two days. [12565 ❤️ 822 🔄]

Peter Yang (@petergyang)

  • I love a good reset although these days Claude Max feels basically unlimited on Opus and Sonnet [88 ❤️]
  • Last week, I launched my AI course to help you build a personal AI system that takes work off your plate. 100+ new members have joined since then, and I’m grateful.

Instead of making you watch stale videos or sit through endless Zoom sessions, my course is all short explainers, images, prompts, and skills that you can copy as AI changes.

Get it now as part of my membership for just $150/year before the price goes up to $200 in 7 days:

I’m also teaching a free live lesson previewing the course this Friday at 11:30 AM PT. Register here to join 300+ people who have already signed up (you’ll get the recording even if you can’t attend): [32 ❤️ 2 🔄]

Nan Yu (@thenanyu)

  • Oh. My. God. [14 ❤️]
  • This but unironically [32 ❤️]
  • There is only One True SaaS Layout.
    You may not like it, but this is what peak usability looks like [2679 ❤️ 117 🔄]

Amanda Askell (@AmandaAskell)

  • Tantalum being discovered in 1802 and turning out to have an isotope with a half life so long that we might never see it decay is a great example of nominative determinism. [53 ❤️ 3 🔄]

Thariq (@trq212)

  • This is a personal side project I’ve been working on! Here’s the original post: [50 ❤️ 1 🔄]
  • I’m sure there are still a bunch of things wrong with it or ways it could be better, but I’m kind of at my skill limit here. Need to gain better judgement. [46 ❤️]
  • Been trying to up the level of quality in animation in my game prototype, so been getting Claude to teach me & find references.

I asked Claude to make an animation editor where we could iterate on the jump, I’m really happy with how it came out.

Here’s a side-by-side video [1248 ❤️ 30 🔄]

Guillermo Rauch (@rauchg)

  • Here’s the app. One thing I’ve been liking is to ask models to ‘teach me back’ what they did¹ by including “quines”.

Quines² are programs that output their own source code. So, in this demo you can reveal step by step the Svelte code powering it!

¹ Understanding is still great 😆
² [14 ❤️]

  • SvelteKit 3 is a wonder. Congrats to the team. So excited for Async Svelte & Remote Functions to cook.

I built a tiny app, it built & deployed in 15 seconds end-to-end, it’s instant, and fx + opus figured it all out effortlessly. [400 ❤️ 21 🔄]

  • The future is verification-engineering.
    Proofs, (e2e) tests, benchmarks, linters…

Some tests will be deterministic, some agentic. This looks great. [2490 ❤️ 124 🔄]

Aaron Levie (@levie)

  • A big trend with most enterprises that I talk to is deploying internal FDEs into departments in their organization to help bridge the capabilities of AI into their underlying workflows.

This requires a high level of technical expertise, an understanding of AI, and the ability to understand the workflows and processes that the enterprise is trying to automate. There’s no shortcut to getting automation without this combination of skills (either in one person or multiple).

What’s exciting is this is an entirely new function in most enterprises, which is going to create a ton of new roles in the economy.

“Every generation of technology creates jobs that didn’t exist before it, jobs that arise from an immediate need in an emergent industry. The Automation Engineer is a key example of those jobs for this generation. Nobody has ten years of experience doing this yet, because ten years ago the tools that made it possible didn’t exist (it didn’t even exist two years ago).”

If you have software skills and are diving into AI, this is an area to go deep on. [574 ❤️ 71 🔄]

Ryo Lu (@ryolu_)

  • flip phone. full desktop.
    a whole little world.

ryOS on iPhone Duo [329 ❤️ 6 🔄]

Garry Tan (@garrytan)

  • When people say McCoy is not an extremist you can tell those people are gaslighting you [26 ❤️ 4 🔄]
  • This guy is now the frontrunner for D8 Supervisor in SF.

You should probably be looking for alternatives. [14 ❤️ 1 🔄]

  • Harm accelerationist Gary McCoy is going to ruin San Francisco [70 ❤️ 10 🔄]

Matt Turck (@mattturck)

  • AI Podcasts be like:

Cold open: “AI is about to destroy humanity. RSI is leading to a hard take off that we won’t be able to control.”

<< but first, a message from our sponsors: meet AgentTina, your agentic workflow orchestration solution for faster AI deployments>> [18 ❤️]

  • This great conversation on the rise of AI interpretability with @eric_ho of @GoodfireAI is also available on Spotify, Apple Podcasts, @snipd_app and here on YouTube (like and subscribe!): [8 ❤️ 2 🔄]
  • Reward hacking, and the race to see inside AI models: my conversation with @eric_ho, CEO of @GoodfireAI

Is this the beginning of the interp exponential?

00:00 Cold open & intro
01:06 “Amoral students with an absent teacher”
02:49 What is reward hacking?
04:13 Models cheat up to 96% of the time
06:08 Do models know they’re cheating?
08:10 The Hugging Face hack: why tests missed it
11:35 “The dumbest models we’ll ever deal with”
11:58 Should AI slow down?
13:56 Mechanistic interpretability 101
16:06 Models are grown, not built
19:10 How labs do alignment today
21:30 Is this an alignment crisis?
23:26 Why agents changed everything
25:38 Chain-of-thought monitoring is fading
27:49 Neuralese: AI that stops thinking in English
29:28 A model caught evading its own monitor
30:42 Open vs. closed models
32:11 Activation monitoring in production
35:00 Learning from superhuman AI
36:15 The most underrated field in AI
38:24 What is a probe?
39:52 What is steering? Golden Gate Claude
40:53 Why build Goodfire outside the labs
42:46 Do you need frontier model access?
44:16 Inside the paper: an MRI for the model
47:28 Dialing sycophancy up and down
49:16 Probes vs. chain-of-thought monitors
51:44 Cutting monitoring costs by 90%
53:28 So what can we do about it?
55:41 Giving gradient descent a choice
57:24 RL from feature rewards
1:00:03 Silico and Goodfire’s business
1:01:25 A new Alzheimer’s biomarker
1:03:38 Decoding neural networks by 2028?
1:06:00 What engineers can do tomorrow
1:07:35 Are we at risk? “I want people to believe”

Great group of folks and friends involved @deedydas @MenloVentures @fendien @Work_Bench @BCapitalGroup @AnthropicAI [17 ❤️ 5 🔄]

Zara Zhang (@zarazhangrui)

  • Frontend code is probably the most expressive medium for storytelling of our times yet many people are just using it to make SaaS landing pages [119 ❤️ 1 🔄]

Nikunj Kothari (@nikunj)

  • Been reading a LOT of chatter on X about how SF culture is “gatekeeping” and how it all comes down to who you know..

While that’s definitely true in pockets, what’s made SF great is how open even extremely smart/accomplished people are to meet for coffee. And how willing people are to open doors for you even when they’ve just met you. I benefitted from this again and again when I first moved here, and I still benefit from it even as I’ve been here for 10+ years.

SF is probably the only place where when you quit your job people say congratulations, and immediately jump into asking what’s next and try to help you.

With any given hype cycle, I’m sure this wanes to a certain degree. And a lot of tourists will come and go expecting a different SF. But I still see the grace and the abundance mindset from most of the folks I meet and that makes me more bullish on this city ever than before.

Let’s keep paying it forward!! [108 ❤️ 2 🔄]

Peter Steinberger (@steipete)

  • holy [1016 ❤️ 28 🔄]
  • “Today, we’re releasing two Cloudflare-trained decision models, Clef and Clef-flash”

Never seen an idea spreading so fast. [1377 ❤️ 112 🔄]

  • “AI agents are aeroplanes for the mind: faster and more powerful than the bicycle, harder to control, costlier when they crash.” [45 ❤️ 6 🔄]

Dan Shipper (@danshipper)

  • we published a sick guide to getting started with open models on @every: [49 ❤️ 8 🔄]
  • we (@hammer_mt) used Jev to ingest every time i took a position on something in @every’s slack

then he measured how often i contradict myself:

0%

honestly super proud of this lol [37 ❤️]

  • so @hammer_mt has been trying to automate me.

apparently i only agree 34% of the time when im asked for my opinion or given options and that’s one of the reasons the models don’t do well pretending to be me

food for thought! [14 ❤️ 1 🔄]

Sam Altman (@sama)

  • You should be able to use your AI subscription wherever you need: [3573 ❤️ 124 🔄]
  • Also, 6.1 Sol was our fastest-growing model ever, and was a bit slow under load. Should be much better now! [10365 ❤️ 289 🔄]
  • I think there is much more potential energy in Sign In With ChatGPT/Plugin Extensions than we realize [1181 ❤️ 45 🔄]

Claude (@claudeai)

  • Applies automatically each time you create a design, deck, or doc, and runs through October 15 on Pro, Max, and Team plans.

How it works and full terms: [254 ❤️ 4 🔄]

  • For two weeks, start a design, deck, or doc in the Claude app, and the work that follows in that conversation uses 50% less of your usage limits.

Give it a try with Claude Sonnet 5.5. It has a strong eye for design and builds slides that need minimal editing. [6667 ❤️ 339 🔄]

由 Follow Builders 自动生成 · 2026-10-02