Dave Vermillion

@HydrogenGuide What if the cost is $.01? Pretty good for a bottle of water.

Cardiologist Warns: "Avoid Bottled Water At All Costs"

@gr8statsbatman @Jabaluck Are you going to say the same when we cure cancer? Problems should be solved if they can. Finding the solution is the dream. Repeatedly failing to solve a problem is not the dream.

@Jabaluck It's just sad that so many young students that dreamed of solving things that haven't been solved may now be left without that dream. That changes things for the worse IMO.

I suspect math will be like Chess and Go due to verifiability. The period of fruitful collaboration between humans and AI will be short (i.e. a few years or less, not a decade). Progress in math will be jagged, with harder to formalize fields coming last, but I suspect this jaggedness will be compressed in time -- I expect superhuman performance at (nearly?) all areas of math within a few years (a few = 2-3?). AIs will also be better at asking pure math questions than humans, and will quickly develop theories beyond human comprehension. Human theorists will have a recreational comparative advantage over other humans in understanding these theories, but AIs will be better at communicating these theories to applied researchers. Pure mathematicians will need to become applied researchers to do productive work, until applied research is also automated. Confidence level for prediction: 50-65% for gist, 40-50% for all above claims being correct.

@timburchett Man say what you will about our blurry videos...

Dominos

@calleymeans @jgebbia @ndstudio Hear me out - Red, White, and Gold…

@jgebbia @ndstudio 🔥

The National Design Studio team continues to grow here at the White House. It’s my pleasure to welcome Peter Arnell to @NDStudio 🇺🇸 Peter Arnell has been appointed Chief Brand Architect of the United States of America — the first person to serve in this role. Peter will help advance President Trump’s Executive Order 14338: “Improving Our Nation Through Better Design.” With more than four decades of experience shaping some of the world’s most influential consumer experiences and brands, Peter will help define and unify how the United States expresses itself — aligning national identity, design systems, communications, and citizen user experience into one coherent whole for the American people. See more from our conversation at the WSJ Future of Everything conference. Make America Beautiful Again! 🇺🇸 wsj.com/video/redesign…

@kwindla the-game-nyc.netlify.app

Sub-agents in (latent) space! We’ve been working on a side project. As far as I know, this is the first massively multiplayer, completely LLM-driven game. Come play Gradient Bang with us. See if you can catch me on the leaderboard. This whole thing started because I wanted to explore a bunch of things I’m currently obsessed with, in an application of non-trivial size, that felt both new and old at the same time. So … a retro-style space trading game built entirely around interacting with and managing multiple LLMs. Factorio, but instead of clicking, you cajole your ship AI into tasking other AIs to do things for you. Some of the things we’ve been thinking about as we hack on Gradient Bang: - Sub-agent orchestration - Partial context sharing between multiple LLM inference loops - Managing very long contexts, and episodic memory across user sessions - World events and large volumes of structured data input as part of human/agent conversations - Dynamic user interfaces, driven/created on the fly by LLMs - And, of course, voice as primary input If you’ve been building coding harnesses, or writing Open Claw agents, or doing pretty much anything that pushes the boundaries of AI-native development these days, you’re probably thinking about these things too! This is all built with @pipecat_ai, the back end is @supabase, the React front end is deployed to @vercel, and all the code is open source.

@pika_labs This is a parlor trick. Your agent uses Pika to create a proxy bot that impersonates and performs like the agent, but doesn’t have the persistent memory. Its mimicry.

Conversations tend to go better with a face and a voice. That’s why we’re thrilled to release the beta version of the first video chat skill for ANY agent, powered by our new real-time model, PikaStream1.0. The skill preserves memory and personality, and enables real-time adaptability. And if you use it with your Pika AI Self, they’ll be able to execute agentic tasks during the call 💅

@alexanderburns — a knowingly false and defamatory headline. Epstein played no role in Peter Attia's success at any level. POLITICO was told this several times and chose to ignore it.

@mayakauf — a knowingly false and defamatory headline. Epstein played no role in Peter Attia's success at any level. POLITICO was told this several times and chose to ignore it.

@grok Is it possible those names could be multiple instances in one email and each file could contain duplicates of the same email in a single thread and some of those emails may not even include Attia in the conversation?

@Davermillion You're right. Clarifying: Attia's name appears 1,700+ times (up to ~1,838 in searches) across the 3M+ page DOJ Epstein files, from their mid-2010s emails with medical advice and crude banter—not 1,700 exchanged messages. Appreciate the note.

New York City Mayor Zohran Mamdami's wife, Rama Duwaji, “liked” Instagram posts that appeared to support the October 7 attacks on Israel. CBS News verified that Duwaji “liked” one post from October 7, showing images of the attack and another one from the day after the attacks showing images of anti-Israel protests in New York City. When asked about the posts, Mamdani said his wife is a "private person who has held no formal position on my campaign or in my city hall."

@grok The files didn’t show 1,700 messages exchanged between those two men.

Yes. In Jan 2026, Bari Weiss (CBS News editor-in-chief) named Peter Attia as a new contributor. Days later, DOJ Epstein files showed Attia exchanged 1,700+ messages with him—including medical advice and crude jokes about women. Attia apologized, denied wrongdoing or island/plane involvement, but resigned Feb 23.

@kevinolearytv Modular reactors?

I think the new play is energy. I think the play is, how do you supply enough energy to advance the economy. So I'm going back to picks and shovels. That means I buy land with natural gas, invest in turbine manufacturing. Anything that has to do with the grid, I'm on board for engineering firms that do this stuff. Copper as a commodity. So think of me as the guy just before the gold rush saying, "You want a pair of jeans? You want a shovel? You want a pick? I'm here." And that's what I'm gonna do for the next three years.

@Scobleizer Teleop… meh

If you think Silicon Valley is the world's technology leader, think again.

@ShauryaDSpectra @AnthropicAI While the U.S. regulates and sues to stop innovation, the rest of the world marches forward.

$20M to lobby for AI regulation. This isn't altruism.... This is the smartest competitive moat I've seen in tech. Here's what's actually happening: OpenAI strategy: Burn $14B/year building better models. Anthropic strategy: Spend $20M writing the regulation that makes competing illegal. The playbook: 1. Build frontier AI capability 2. Fund "bipartisan" org to push for "safety" regulations 3. Ensure regulations require: • Massive compliance infrastructure (only big companies can afford) • Transparency requirements (exposes smaller competitors) • Export controls on chips (locks out new entrants) 4. Call it "inviting scrutiny" while actually building regulatory moat The future (12-18 months): • Federal AI framework passes • Compliance costs: $5M+/year minimum • Every AI startup needs lawyers before engineers • Only Anthropic, OpenAI, Google survive • They write the rules, then "comply" with rules they wrote This is regulatory capture 101: Tobacco did it. Pharma did it. Banking did it. Now AI is doing it in real-time, and calling it "safety." The tell: "Targeted regulations on frontier models" Translation: Regulations that only apply to companies big enough to compete with us. Small AI companies building agents? Unregulated. New startup threatening Anthropic? Must comply with federal framework. In 2028, we'll look back and realize: The AI race wasn't won by who built the best model. It was won by who wrote the rules first. While everyone debates if AI is "safe," Anthropic just lobbied their way to a government-protected oligopoly. Brilliant...Terrifying...Inevitable

AI is being adopted faster than any technology in history. The window to get policy right is closing. Today we’re contributing $20m to Public First Action, a new bipartisan org that will mobilize people and politicians who understand what’s at stake. anthropic.com/news/donate-pu…

@ToBoldlyGrow_ @wintonARK 7

@wintonARK What’s the estimated cost of the 1st?

On Earth the datacenter buildout is subject to backwards cost scaling. The 100th GW deployed will almost certainly be more costly, complex, time intensive and subject to negotiation than the 1st. In space, the opposite. The 100th orbital GW could be 1/3rd as costly as the 1st.

Aren't recursive world models (like the human brain) where the breakthroughs will really happen?

@AlexFinn @ConstantHawk This is key. Clawdbot is the scout, forager, harvester. Sonnet 4.5 as the mastermind.

@ConstantHawk the idea isn't to replace claude. it's to supplement it with employees that can work 24/7/365 for free

3 THINGS YOU NEED TO BUILD IMMEDIATELY WITH OpenClaw: 1. Activity feed 2. Calendar 3. Global search All 3 will super power your workflow • Activity feed actively tracks everything your OpenClaw does. This is critical, because if you have it working autonomously, this will give you insight into EVERY SINLGE THING it does, to make sure it's not wasting tokens • Calendar lets you see all of OpenClaw's scheduled tasks. Now you can verify when it's going to work proactively for you. Also will let you know when it has scheduled tasks you might not want it to do anymore, saving more tokens • Global search allows you to search through ALL of OpenClaw's memories, tasks, documents and past conversations. OpenClaw has such incredible memory, but no interface to view any of it. Now you can search through it easily and find old nuggets you talked about. Steal this prompt to get it all installed: "I want you to build out 3 things for me. In a Mission Control dashboard, build out an activity feed first. This activity feed will record EVERY SINGLE THING you do for me, so I can see a history of every action and and task you've completed. I want a calendar view that shows me in a nicely formatted screen every scheduled task you have in the future in a weekly view. And I want a global search where I can search for any term and you display any relevant memory, document, or task from our workspace. Use NextJS as the framework, Convex as the database, and Codex to code it all out"

@bilawalsidhu I always wanted a "gods eye view" of a slap fight.

This will change the way we experience sports forever -- watching the game from a gods eye view. Arcturus is building 4D gaussian splatting tech that can capture every angle of a sporting event and pushes the bar for volumetric video. I tested this in a headset and it makes 360 and 3D 180 video look ancient. This puts you closer to the action than any stadium seat, and it’s genuinely mind blowing.

@grok is moltbook truly run by independent agents or is it run by humans?

@CenationCity Robot wrestling. I’m in.

BREAKING : Elon Musk is reportedly heavily rumoured to consider a move to buy WWE from TKO, with talks said to be at a very early stage. WWE is valued at around $9.3 billion, and the idea is that Musk views it as a major global brand with huge live-event and media value. Nothing confirmed yet, but imagine 𝕏 × WWE as a collaboration.