For the first time in history, GPT-6 Astra created a fly that is simply a fly. Not a chatbot pretending to be a fly. Not a simplified animation. A digital fly running its own connectome brain, perceiving a real room through its own senses and interacting with the environment using the power of AR. The entire loop is surprisingly wild: Spectacles capture the retina, light rays, hands, and the surrounding room → the room is converted into an inventory → sensory information is fed into the fly’s neural system → the MaleCNS connectome processes it → descending neurons and motor pools generate actions → the commands return to the Spectacles → wings, legs, and head respond. In other words, the fly sees through its own eyes, processes the world through its own simulated brain, and acts through its own virtual body. We are moving from AI that generates outputs to AI that actually has a body, senses, a brain, and a closed perception-action loop. The 🪰 is no longer just simulated. It is experiencing a world The developer @pt_pavlo is pinned below in the post

GPT-6 ASTRA DID IN 7 DAYS WHAT SCIENTISTS SPENT 20 YEARS MAPPING. Scientists spent two decades reconstructing the brain of a fruit fly, and now AI has taken that map and turned it into something that actually works. The fly has just 166,700 neurons, packed into a brain roughly the size of a poppy seed, yet its digital counterpart can already generate surprisingly complex behavior, raising an uncomfortable question about how much computation we really need to reproduce intelligence. Most AI systems learn by consuming enormous datasets and running billions of computations until useful patterns emerge, but this approach takes a completely different path: instead of training a giant model on the world, researchers took the physical wiring of a biological brain and simulated it directly inside a computer. - No giant dataset. - No traditional training. - No billions of learned parameters. The digital fruit fly reached roughly 91% behavioral accuracy using only four ingredients: the complete neural connectivity map, synaptic connection counts used as connection weights, the classification of neurons as excitatory or inhibitory, and an extremely simple Leaky Integrate-and-Fire neuron model. That is the part that should make you stop and think. The intelligence was not explicitly programmed into the system; much of the behavior emerged from the structure itself, suggesting that the topology of a biological neural network may encode far more of the brain’s “software” than we previously understood. A model with fewer parameters than most modern AI systems can already do something remarkable because the fly’s brain has already been shaped by evolution for millions of years. Your LLM has more parameters than there are flies on Earth. The fly already shipped. And if the same idea can scale from 166,700 neurons in a fruit fly to millions in a mouse and eventually toward the roughly 86 billion neurons in a human brain, whole-brain simulation could become one of the most important frontiers in AI. This is not the fly becoming human. This is you realizing how little you needed