Nick

LIFT is offering public flights in its personal electric aircraft over Lady Bird Lake in downtown Austin. Now available to ride for $199.

Hardware is the moat that buys you time to build the software castle.

At first I didn't realize how big this was, until I read the fine print: TLDR with the insights baked in: OpenAI has developed a novel language model called GPT-4b micro, specifically designed for protein engineering in collaboration with Retro Biosciences, a longevity research company backed by Sam Altman's $180 million investment. The model demonstrates remarkable capability in optimizing Yamanaka factors, which are proteins responsible for cellular reprogramming from regular cells to pluripotent stem cells. The model has achieved a significant breakthrough by engineering proteins that are 50 times more effective at cell reprogramming than existing versions, with success rates surpassing human-designed alternatives. This specialized small language model differs from AlphaFold by focusing on sequence manipulation rather than protein structure prediction, utilizing a training dataset comprising protein sequences across multiple species and protein interaction data. The model employs a few-shot prompting methodology and can suggest extensive protein modifications, altering up to one-third of amino acids in target proteins. While the collaboration between OpenAI and Retro Biosciences was conducted without financial exchange, it has raised questions about potential conflicts of interest due to Altman's dual role as OpenAI CEO and Retro's primary investor. The results await peer review and publication, and the model remains a demonstration rather than a commercial product. Now, let me explain it simply, and why token-reading transformers combined with other narrow-AI solutions like AlphaFold will be a total gamechanger: What makes this absolutely revolutionary is how perfectly suited transformer models are for this exact problem. Think about it - transformers process sequences of tokens, and what's DNA? It's literally nature's token sequence. Proteins are just strands of DNA wearing a fancy outfit, and transformers can read them like a master cryptographer breaking the simplest cipher. This is where it gets interesting. Transformers don't just read the sequence - they understand the abstract patterns, the higher-level purpose, the whole evolutionary story behind why that protein exists in the first place. When you combine this with structural prediction tools like AlphaFold, you're not just reading the blueprint - you're understanding the architect's entire vision. The medical implications are staggering. We're talking about a future where we can read your genome, understand every protein in your body, and engineer precise solutions for everything from cancer to aging. This isn't science fiction anymore - it's engineering, and it's happening right now. But here's the kicker that nobody seems to be talking about: this is a "micro" transformer. We're not talking about some massive supercomputer - this could run on your smartphone. Right now. Today. Remember Star Trek's medical tricorder? That's not just fantasy anymore - it's the logical conclusion of where this technology is headed. OpenAI isn't just building another AI model here - they're laying the groundwork for a complete revolution in medicine. And the best part? We're just getting started. This is version one of a micro model, and it's already outperforming human experts by 50x. Imagine where we'll be in five years. Welcome to the future of medicine. It's going to be a wild ride.

🚨 NEW: Marc Andreessen on China's manufacturing dominance "There's three industries that follow phones that the Chinese own the global market at: 1) Drones Something over 90% of all the consumer drones are made in China. Which is what the US Military also uses. It's the whole ecosystem, it's all of the componentry. The Biden FAA has been trying to kill the US drone industry as part of its war on tech. China is now sanctioning our own drone companies for batteries made in China. They have leverage. 2) Cars The drone supply chain is analogous to the car supply chain. The new Chinese cars that are coming out are really good and they have a giant advantage on cost 3) Robots The Chinese are 100% of us in Robots. We have the leading R&D but they have almost complete dominance in manufacturing capability. They have the 1000s of companies that make all the components that go in robots. Our Boston Dynamics SPOT robot cost $50,000. China has a similar product that looks and acts similar and is even hooked up to an LLM and it only costs $1,500. This is all upstream from all the military applications because it's all the same supply chain. We have to confront this. Having this fragmented approach where we hate tech and then somehow think we can develop a competitive response to tech is not coherent. We have no whole of government strategy on technology or China. Zero. It turns out the President matters.