R
ARCHIVE
内容归档
正文、图片和 Markdown 均已本地保存，不包含回复区。
G
A
Mark Pincus, founder of @Zynga , on "proven, better, new," the product framework that was Zynga's religion: 1. Proven: start with what 10/10 existing users of a category already love, and don't change it. "You do not have the right to change proven. You have not taken enough time to change proven. It is an art that we do not completely understand." 2. Better: only add what 10/10 users agree is better. "For Zynga Poker, better was no download. That was it. Turns out you lose half your users every time you click download." 3. New: treat anything you think is better as new, and assume it fails. "Assume all new fails. Write that on the wall... If all new worked, we'd be using new s**t all the time. But how often do you change what's on the front of your iPhone?" "Users don't care how innovative it is. All they care about is does it feel better, and they can't even tell you why." @markpinc w/ @JoshLu 🔗 View on Twitter speedrun @speedrun SR008 APPS ARE OPEN. Up to $1M + $10M+ in credits. 12 weeks in SF, Feb through April. Built to help you land capital, customers and talent. Apply by Nov 1: speedrun.a16z.com/apply 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
V
It's remarkably short-sighted for websites to block AI agents. Agents WILL become the way most people transact online - block them and you'll lose a ton of customers. And I suspect LLMs will get better at browser use faster than humans will improve at detecting them... Wise @trikcode notice.. > Delta is blocking AI agents from booking flights > JPMorgan Chase and Charles Schwab block agents from accessing accounts > Yelp requires paid data licensing for agents > Amazon has blocked Grok bot and other shopping agents. the internet is starting to block AI agents that were supposed to do everything for us 🔗 View Quoted Tweet
S
🎯 Michael Seibel @mwseibel The SF homelessness trap, simply: Advocates convince liberals they're helping locals priced out of housing. The city spends billions on housing and services. But many recipients are drug users from outside SF. Users attract dealers. Drugs stay cheap, services stay generous, and more people come. Cartels see this and bring in cheap fentanyl. Problem explodes. Advocates mean well, but the unintended consequences are too great. The money we spend on housing, treatment, and healthcare for out-of-town drug users could go to San Francisco Kids & Schools. 🔗 View Quoted Tweet
S
P
Camera Director adds a cinematic camera move to any clip, from an orbit to a slow push-in, right inside ChatGPT. The performance stays exactly as it was. 🔗 View on Twitter Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
R
F
T
P
Film the dance once and Motion Transfer teaches it to any character, right from ChatGPT. No dancing skills required. 🔗 View on Twitter Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
L
Exciting news: DeepSeek-V4.1-Flash-20261017 (Max) by @deepseek_ai is now the #1 open model in Code Arena: WebDev, surpassing Kimi K3 (Max) at 10x lower cost! Highlights: - #1 open model, ahead of Kimi K3 (Max) - #12 overall with 1670 pts, ahead of Claude Opus 5 (High) and within ~10 pts of GPT-6 Sol (Max) - Shifts the Pareto frontier at $0.3/$1.2 per million input/output tokens - +52 pts vs. the previous DeepSeek-V4.1-Flash-20260910 Congrats to the @deepseek_ai team on another strong contribution to the open source ecosystem! DeepSeek @deepseek_ai 🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness , Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4 🔗 View Quoted Tweet
L
L
Anthropic AI researcher: “A lot of what we do feels so much closer to biology than to engineering.” 🔗 View on Twitter Lenny Rachitsky @lennysan "To make them is not to understand them." Emmanuel Ameisen ( @mlpowered ) has spent the last 3 years doing interpretability research at @AnthropicAI , looking inside LLMs like Claude to figure out what they're actually doing. As he puts it, models are "grown," not built. We discuss: 🔸 His discovery that models plan many tokens ahead 🔸 Why anthropomorphizing AI can actually be the right approach 🔸 What the "J-space" reveals about how models think 🔸 How interpretability connects to AI alignment and safety Listen no youtu.be/JHo-aiePjoM fpkab 🔗 View Quoted Tweet
含视频地址
A
Aaron Levie of @Box on why AI widens the gap between experts and everyone else, and what that means for college students: "The biggest gains of AI go to people who have some degree of expertise in an area, to know what is actually true, what is not gonna work, 'What should I integrate from the output of this AI?'" "If you don't have a deep understanding of your particular space or field or domain, you aren't able to then have the right judgment to make all of those decisions." "So I think the experts just get more powerful in this world." "I'm not even convinced that you can tell a college student to learn anything different than any other period in history. Be really good at a particular field, and then AI is merely a turbocharger of your capability in that particular field." @levie 🔗 View on Twitter a16z @a16z Jad Naous of @grepr_ai on how AI is widening the gap between okay and great engineers: "Some people were adapting and learning and others weren't as adaptive. And then as we started hiring more people into the team, that disparity between okay engineers and really good engineers just multiplied 10X." "Ultimately you really have to understand the setup. The more complexity that your brain can handle, the more you can break down the tasks and assign them." "My team is still small... In my experience, for the things that we do, that is the case because we build distributed systems and distributed algorithms, and your brain has to handle the complexity because AI doesn't do it." "For the problems that we solve, none of the models can actually think through the things that we do." @jadtnaous @martin_casado 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
J
One of the most important challenges in advancing enterprise AI is creating evaluation benchmarks that reflect their real data and workflows - without using their actual data or workflows due to privacy/security restrictions We see our customers upload a massive long tail of complex documents for extraction, which break even the most complex frontier models. In order to create a representative benchmark that adheres to this complexity, we invested heavily in a synthetic pipeline that can mine real document templates, but allow flexibility for variation of not only the data, but the template and rendering. Come check out ExtractBench Pro! llamaindex.ai/blog/introduci… Website: extractbenchpro.ai 🔗 View on Twitter Jerry Liu @jerryjliu0 Introducing ExtractBenchPro: the most challenging benchmark for real-world document extraction. We see a massive long tail of document complexity in terms of use cases brought on from our customers, that are completely unsolved from a price-performance standpoint. These include trial balances, well logs, canvass binders, site plan reviews, duty reconciliations, loss runs, and more. Most existing extraction benchmarks are saturated by the frontier models (our cheapest modes now score 93%+ on ExtractBench). So there is enormous economic value in building a complex extraction benchmark that reflects this long-tail of knowledge work. We spent a long time curating the most complex elements that we saw from real world extraction use cases: ✅ Document conventions: containing legends or rules where the rest of the document should be interpreted ✅ Density of internal jargon: specific definitions that must be cross-referenced within a doc ✅ Aggregation and reasoning: beyond returning exact values on a page, infer values derived from aggregating multiple sources of information ✅ Drawn relations: complex diagrams like procedures and circuits ✅ Charts inside documents: read precise points within a chart ✅ Long chart sheets: some data series like well logs cram all this data into an extremely long page Within the industries where document extraction is relevant, even small mistakes can have enormous negative impacts. Therefore our headline evaluation metric is based on perfect documents: the proportion of docs with perfect (100%) value f1 score. (Besides this we also measure value f1 and grounded f1) It’s an extremely hard task for even the latest frontier models to solve, topping out at ~58% or so in Value F1 score for Codex with Astra - which is quite good but extremely expensive (almost $1 per page!) Our LlamaParse agentic/agentic plus modes are 20-40% and cover the left side of the Pareto frontier, at price ranges that are feasible for full production use. ExtractBenchPro is available today as a website/technical blog, and also on Github and HuggingFace. Check it out! Blog: llamaindex.ai/blog/introduci… Website: extractbenchpro.ai Hugging Face: huggingface.co/datasets/llama… Github (is ExtractBench): github.com/run-llama/Extr… 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
G
A
We’re committing a portion of Anthropic’s shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit. It will start off with 5% of our shares. For every 50% increase in the company’s value over a set baseline, we can contribute up to an extra 2%, subject to our board’s discretion. In total, this means that roughly 20% of Anthropic’s shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program we’ve run since our founding. In the post below, we share more details about our new fund. anthropic.com/news/public-be…
F
Y
I'm definitely not anti-achievement and never criticized real achievements, including by Elon. I even own two Teslas. I do expose marketing stunts that, while cool on the surface, are not the achievements they are claimed to be. Compressed air thrusters might give you acceleration boosts in any direction. They might even make your car hover for a couple of seconds if you accept replacing the passenger seat by a 400 liter high-pressure air tank. But they are not a path to actual flying cars. The same way Burt Rutan's SpaceshipOne was never going to be a path to orbital flights (just bunny hops out of the atmosphere). I am anti-Musk when it comes to politics. His political actions, the massive dissemination of disinformation through X and other channels, and his support for authoritarian regimes and racist ideas have the effect of weakening democracy and weakening human decency and social progress around the world.
S
If you can use models to reverse engineer anything and use models to swarm and find vulnerabilities, we are not ready for what's going to happen over the next 12 months. We've now have the autonomous black hat hacker. Tremendous crisis right in front of us. Suhail @Suhail One thing to consider is that the best hacks on American companies will go down very quietly. A great hacker is unseen. So you should presume that attacks on American companies that are noticed, are a small portion of what’s happening out there. Most hacks are just silent and steadily waiting to accumulate more valuable information on an on going basis. 🔗 View Quoted Tweet
G
A
P
Ask ChatGPT for a mascot and Character Studio delivers a full reference sheet, and saves them to your library, ready for their next shoot. Pika @pika_labs Creative work demands a visual environment. So we partnered with OpenAI to turn ChatGPT into one. Introducing the Pika X ChatGPT plugin extension. All of Pika’s creative tools are now directly in your chat and side panel, in a new interactive creation experience. More on what’s new 👇 🔗 View on Twitter 🔗 View Quoted Tweet
含视频地址
O
// Latent Semantic Memory for LLMs // This paper presents some really interesting findings about agent memory. They find that sixteen memory tokens beat full-context inference on FactKG while using 108x fewer tokens for the source documents. Approach: Documents are encoded once into reusable embeddings. For each query, a learned module gathers the relevant information from the whole collection and writes it into 16 tokens in the input space of a frozen LLM. Results: It beats Mem0, MemoRAG, and other baselines on five of eight benchmarks. On FactKG, it scores 84.33% against 80.43% with the full context. It also works as agent memory. In TextWorld, using the agent's own history as the collection raises success from 18.6% to 24.2%, and the agent revisits fewer rooms. It also seems that it could help frozen LLMs accumulate and use knowledge effectively across long-context interactions. This is applicable to building continually learning AI systems. arxiv.org/abs/2610.21375 Chat with Paper: academy.dair.ai/papers/memoril…
F
Supabase怎么做到100亿美元？5分钟讲透 (in Chinese) How did Supabase reach a $10B valuation? 1M+ new databases a week, ~70% created by AI tools. Not just luck: AI coding tools picked it as the default (one "Integrations" button in v0 doubled new databases). But that default sits in other people's hands: v0 later switched to a rival. Sources: Supabase blog & press release, CEO Paul Copplestone's posts, TechCrunch, npm. Not investment advice. 🔗 View on Twitter
含视频地址
V
F
The animated cut (2:41, no narration, in Chinese) of Alan Walker's essay on Musk's line: "The future light cone of discoveries rounds up to 100% engineering." ~26 parameters in the rulebook vs 10^130 possible proteins, a 48-year wait for the Higgs, and AI predicting 2.2M crystals at once: verification becomes the bottleneck. All code, no footage. 🔗 View on Twitter
含视频地址
R
This is so cool. One unified view of all your agent sessions. Across personal & coding agents, apps, API keys. Vercel AI Gateway helps you run your business better. You own the model choice, your data (ZDR), and crucially the *meta*data to understand RoI. Vercel Developers @vercel_dev See every agent session in AI Gateway logs. Sessions group requests by conversation, with cost, cache hit rate, and the agent that ran it. Session affinity keeps each one on the provider holding its prompt cache. vercel.com/changelog/ai-g… 🔗 View Quoted Tweet
F