Pinned Tweet
Mark Watney had to science the sh*t out of Mars.
An AI should have helped him.
One that could look at a broken machine, tell him how to fix it, and show him his options with what he had.
That's what I'm building. Starting with a vision app that looks at any machine and tells you what it is and how to fix it.
You solve one problem, then the next, then the next. Solve enough and maybe it ends up on Mars.
Ben retweeted
Last night an AI agent performed brain surgery on a mouse.
We gave it 19 gene therapy vectors in a spreadsheet, a mouse model of epilepsy, access to a motorize surgery rig with cameras and one goal: reduce seizures.
It picked the drug. It picked the brain region. It derived the coordinates from three papers. Then it drove the stereotaxic arms into the brain and injected 250 nanolitres while watching the entry site through the rig camera for bleeding.
A licensed vet gated every irreversible step and held a physical kill switch to stop the AI agent.
Thread on what it decided, what it saw, and the moment it refused to let me override it.
[Writing this on the spot out of excitement, but I've been 18hrs in front of my laptop coding and coordinating this, so keeping it brief and descriptive, more cohesive thoughts on this soon...]
Ben retweeted
Excited to share Genie 4 today our most realistic and interactive world model:
- steer dynamic environments in real time, shaping events, camera angles, and actions
- generate native audio matched to the world, from realistic background sounds to characters who can converse in real time!
- simulate photorealistic scenes and real-world physics
- create more coherent multi-step events through reasoning from Gemini
We've started to roll out as a research preview to trusted testers. Coming soon to Google AI Ultra subscribers in Project Genie and early access partners via the Gemini API and AI Studio.
Ben retweeted
JUST IN: Google DeepMind unveils Genie 4, a new AI model that generates photorealistic, interactive 3D worlds in real time, complete with physics, sound, & human dialogue.
“It’s not enough that you believe what you see. You must also understand what you see.”
- Leonardo da Vinci
Ben retweeted
there's something genuinely beautiful about topology optimization
you give it a load case, a boundary, and a material budget and it just… removes everything that doesn't need to be there
no aesthetic intent
no design language
no human bias toward familiar shapes
just physics deciding what stays
the results look organic because nature runs the same algorithm stress flows material remains where it must everything else disappears
engineers spent centuries adding material for confidence
topology optimization asks a different question entirely:
what's the minimum structure that doesn't fail?
and the answer is always something that looks alive
Agents will need a todo list
Will start with Trello
Create a card for 1 task and connect it to the agent
I’ve been using agents wrong.
I’ll be digging deeper into it, sharing learnings and ways to orchestrate better.
How is everyone using it? Tips always appreciated.
Does it hold up for at least a century?
Those tests and simulations will be interesting to see
Great new developments but the question is if they can hold up over years
Nature is quite good at decaying things
