To people who think "China is surpassing the US in AI" the correct thought is "Open source models are surpassing closed ones" See ⬇️⬇️⬇️

The new OpenAI paper about “why language models hallucinate” addresses the most important problem with LLMs! Hallucinations are simply guessing errors Our benchmarks and evals reward guessing because you get more answers right that way This translates to LLM training

[2506.02153v1] Small Language Models are the Future of Agentic AI arxiv.org/abs/2506.02153

We cannot understand the true nature of the Universe, unless we question deeply. I want to know what is real, even if the answer is total obliteration of my consciousness.

.Marc Andreessen 🇺🇸 on drive:

Fei-Fei Li (Fei-Fei Li) on limitations of LLMs. "There's no language out there in nature. You don't go out in nature and there's words written in the sky for you.. There is a 3D world that follows laws of physics." Language is purely generated signal.

Videos show Nepal’s parliament on fire as protests continue, with PM Khadga Prasad Sharma Oli and other top ministers resigning over mass anti-corruption unrest that left at least 19 people dead.

.Chris Dixon: The next big thing will start out looking like a toy (2010)

Ray Dalio’s reply on Reddit when asked what’s the most important decision you can make

Measuring growth in crypto: What’s different, what matters, and what needs to be adapted.

Fei-Fei Li says language models are extremely limited. This Google DeepMind paper makes almost the same point, just in the world of video. The models are just very advanced pattern matchers. They can recreate what looks like reality because they’ve seen so much data, but they

Wall Street closes. Solana never sleeps. Huge thanks to everyone who joined us at the All-In Summit 🙌

The life of a (tired) show girl in 2025

.Marc Andreessen 🇺🇸 on why initial startup ideas are overrated:

Well done!

I'm bullish on founders who choose to raise smaller seed rounds, even when they have the option of a bigger round out of the gate. the #1 reason - smaller rounds encourage FOCUS early on in a start-up, focus is so important - streamlining the team around a single target customer

If you reach your goals by compromising your values, you haven’t succeeded. If you fall short of your goals by upholding your values, you haven’t failed. The true measure of accomplishment is standing by your principles when they’re tested.

Holy shit... Tencent researchers just killed fine-tuning AND reinforcement learning in one shot 😳 They call it Training-Free GRPO (Group Relative Policy Optimization). Instead of updating weights, the model literally learns from 'its own experiences' like an evolving memory

Good cofounder > no cofounder > bad cofounder. This may seem obvious but a lot of people don't seem to grasp it.