David Webster

imagination, not intelligence, made us human

@a16z @vishalmisra this was my idea: x.com/webstardavid/s…

The Einstein Test for AGI. @vishalmisra: “You take an LLM and train it on pre-1916 physics and see if it can come up with the theory of relativity.” “If it does, then we have AGI.” “It’s a high bar, but we should have high bars.” “At the time of Einstein, there were a lot of clues that Newtonian mechanics, there was something missing.” “But until Einstein came up with a new representation of the spacetime continuum, we were stuck.” “If you had a model that just looked at correlations and such... it would not have come up with the beautiful equation that Einstein came up with.” Vishal Misra with @martin_casado

@Mercantio @nvidia i like graveyard shifts

@WebstarDavid @nvidia A family father shall go to sleep no later than 00, and no earlier than 21. It's the law

playing with @nvidia nemoclaw $2 for free :P she is alive!!! nvidia.com/en-us/ai/nemoc…

@Mercantio @nvidia technically its work prospecting while i fix some broken stuff :/

@WebstarDavid @nvidia Remove it just a shell with some random patterns and an LLM on top and go sleep boss

@nvidia running nemotron-120b at $0.18 == 10 free hours

@simonw we are watching the clawpocalypse

Claude CoClaw

@cmuptx sexy af

🚀 Excited to release AIBuildAI — an AI agent that automatically builds AI models 🏆 #1 on OpenAI MLE-Bench 💻 GitHub: github.com/aibuildai/AI-B… Building AI models still requires a lot of manual work — designing models, writing code, training them, tuning hyperparameters, and iterating on results. We developed AIBuildAI to automate much of this process. The system runs an agent loop that automatically: • analyzes the task • designs models • writes code to implement them • trains the models • tunes hyperparameters • evaluates model performance • iteratively improves the models 🏆 On OpenAI’s MLE-Bench benchmark, AIBuildAI ranked #1: github.com/openai/mle-ben…

@AndrewYang good or bad?

👀

@deedydas holy wow thats 15 zeros congrats

OpenRouter just broke 1 quadrillion tokens a year. Assuming ~$1/M, $1B would be spent on it annually.

i always like james gosling the creator of java, say what u want its hard to engineer something that backwards compatible. also his brutal take on genai being a science experiment.

@mark_k but when i hit that sweet hyper focus and go till 4am that is sweet

@WebstarDavid Me too tbh

Research shows that ADHD brains can slip into brief episodes of "sleep-like" activity during demanding tasks, even when the person is fully awake. This surprising neurological pattern directly impacts focus and daily functioning. These momentary micro-lapses help explain many classic ADHD struggles. When the brain drops into this state, it naturally leads to slower reaction times, increased mistakes, and sudden drops in concentration. Findings like this offer profound validation. They provide a clear reminder that struggling to maintain attention is not a lack of willpower or effort. It is simply a brain temporarily operating on a different wavelength.

@DeryaTR_ haha good point, they are probably the most resistant to ai

@WebstarDavid Some hunter-gatherers were really really good too ;)

Those still at L1 (No AI) are like the hunter-gatherer level of coding at this point 😅

@mustache_dev @threejs hawt

I built a physically based @threejs drifting simulator. Let me explain 👇 #threejs #gamedev

@KevinNaughtonJr we are 6 months away from september

we are 6 months away from AI replacing all software engineers

@TukiFromKL wow crazy, should pay real creators more

🚨 Did you just see what YouTube did? YouTube isn't banning AI slop.. They're making you label it so they can train their next model to not look like slop. Read that again... You flag the bad AI content. YouTube collects it. Google feeds it into Veo 4... Then next year their AI generates videos so good you can't tell the difference... You're not the customer... You're the training data, we all are.. They got millions of creators to do free quality control on their next AI model and convinced them it was "a strong stance."

@OpenAI i never use the mini models i need the best

GPT-5.4 mini is available today in ChatGPT, Codex, and the API. Optimized for coding, computer use, multimodal understanding, and subagents. And it’s 2x faster than GPT-5 mini. openai.com/index/introduc…

@kirawontmiss wonder how he feels about moss

never in my life did i think i would live to see the day someone starts hating on trees

I hate it that the browser is a better calculator than the calculator app

@dwarkesh_sp nvidia must be loving this

The value produced by models is getting so much better so fast that old hardware is actually getting *more* expensive to rent. 3 years ago, the best model you could run on a H100 chip was GPT-4. Now, you can run GPT-5.4 on it, which is smaller and cheaper to run while producing much more valuable tokens. w. @dylan522p