A fly that died years ago is flying around a harbor that doesn't exist. Here's what's going on. Scientists took one female fruit fly and sliced its brain into about 7,000 pieces, each roughly 2,000 times thinner than a human hair. They photographed those slices 21 million times under an electron microscope. Then hundreds of people spent years tracing every connection by hand until they had the whole thing, all 139,255 neurons and 54.5 million synapses. That map is called FlyWire. Odyssey hasn't said exactly which map it used, but FlyWire is the one most of these projects build on. Earlier this year a startup called Eon ran that map as a simulation with zero training and got a virtual fly that walks and grooms on its own. Now Odyssey has put the brain somewhere new. The harbor in this clip isn't sitting on a server as a 3D model or a game level. Odyssey-3 is a 14-billion-parameter AI that learned how water and light behave by watching a ton of video, and it invents each frame as the fly steers through it. This is kind of a big deal for a nerdy reason. For as long as people have tried to build a mind, there have been two camps. One says copy a real brain. The other says train a giant model on tons of data. Those camps have mostly fought each other, and in this clip one is living inside the other. And here's a fun bit of math. The fly has 54.5 million synapses. The AI dreaming up its world has 14 billion parameters. Line those up and the world is about 250 times bigger than the brain living in it. That's what happens when one approach rides on GPUs and the other rides on grad students with microscopes. The trained side ballooned in a couple of years. The copied side took the better part of a decade to get one fly. What really got me is how little it takes to fool a fly. A fruit fly sees through about 800 tiny lenses per eye, so its whole view of the world is around 1,600 pixels. Each of your eyes sends about a million nerve fibers to your brain. Keeping a fly convinced means the AI only has to look right across 1,600 pixels, which is a pretty low bar, and it's clearing it. The fake world will look good enough for human eyes long before anybody can slice up a human brain and rebuild all 86 billion neurons.
We placed the fruit fly brain inside Odyssey-3, a frontier world model. The lil’ guy is now flying around totally imagined, simulated worlds all by itself 🫶

Oct 19, 2026 · 8:14 AM UTC

55
169
1,718
384,372
Sort replies: Relevant Recent Liked
Replying to @aakashgupta
A little while later ...
Made with AI
2
84
2,842
Replying to @aakashgupta
We're the fly
14
2,701
Replying to @aakashgupta
A fly flying around a harbor that doesn't exist, with no idea anything is off. Some mornings that's the most relatable post on here.
32
3,614
Replying to @aakashgupta
Fruit Fly Consciousness? This whole exercise is fascinating. And it seems to beg the question of whether a fruit fly (even a meticulously mimicked one) can be given a consciousness in an imagined world of its own making? This assumption is wrong. A LLM is only taking the data it was given and predicting what should follow. Even a model with 14 billion parameters is only predicting (albeit better than predecessors) what the mapped out brain of a fruit fly could see, based on its training. It seems natural the mimicked fruit fly would be shown to be flying in a made up world, predicted (not imagined) by the model and NOT the fruit fly. To put it another way, the fruit fly does not exist. Therefore, the LLM is not fooling something that isn’t really there. So the represented image of it is not one of an actual “virtual” fly, imagining flight over an imagined harbor. It’s just the prediction of what a fruit fly’s flight might look like.
1
266
Replying to @aakashgupta
You didn’t fool a fly. You wrote a computer program to mimic a fly brain and then had that computer program “fly” inside another computer program.
30
Replying to @aakashgupta
And this fly does nothing but 'fly'..finds no mate, rotten meat, people to annoy, dogshit, bin? Seems like any fly you know?
9
1,722
Replying to @aakashgupta
It's great that the fly can see where it's going but can we trust that it is doing what it wants to do without any other sensory inputs such as smell or taste? will it go insane without these inputs?
10
Replying to @aakashgupta
I don't know about the fruit fly, but the boats in the harbor are honestly terrible. Many of the boats make absolutely no sense. This AI model may understand water, light and flies. But it doesn't understand boats.
8
Replying to @aakashgupta
Is anyone studying the ethics of this?
6
1,529
Replying to @aakashgupta
Is the image what the fly sees or what its brain interprets. What controls the focus on distance instead of near?
1
1
2,296
Replying to @aakashgupta
Welcome to The Matrix
63
Replying to @aakashgupta
Is the fly required to watch adds for additional tokens?
Replying to @aakashgupta
Who is to say that this isn’t already the “reality” we live in?
32
Replying to @aakashgupta
Might want to read Daniel Suarez Kill Decision. Also Daemon as a base.
1
1,159
Replying to @aakashgupta
great map. now show me which neuron makes it check-raise.
23
Replying to @aakashgupta
> The fake world will look good enough for human eyes long before anybody can slice up a human brain and rebuild all 86 billion neurons. How much would that cost? Probably much less than one data center. We should do it.
1
1
1,510
Replying to @aakashgupta
Does it autonomously land in dog shit?
437
Replying to @aakashgupta
😃 now let’s see a kitchen flyswatter battle . . .
494
Replying to @aakashgupta
its own little view in the corner looks less impressed than the rest of us
1,925
Replying to @aakashgupta
But a fruit fly in my house doesn’t fly that much or that far. And it sits a lot.
1
519
Replying to @aakashgupta
Would the fly be conscious on multiple planes of existence then? @grok @bot proof of multiverse? Or is this study bullshit
1
204
Replying to @aakashgupta
This is a silly high school science project.
33
Replying to @aakashgupta
without validated neural dynamics, fly brain wiring in odyssey-3 isn't proof of a recreated brain
2
203
Replying to @aakashgupta
This guy ran 20,000 fruit fly brains at once. 👀
Kevin D. Johnson | Heterogeneous Compute Ontology
39
Replying to @aakashgupta
Sorry. Could not understand even a bit..
1
Replying to @aakashgupta
Why would we want to do that
54
Replying to @aakashgupta
The problem is that if you look at the model built, it’s frozen in time… very fascinating for readers but still far from being exactly as claimed… VERY FAR
1
4
723
Replying to @aakashgupta
It is fastest fruit fly I have ever seen.
77
Replying to @aakashgupta
Is it conscious
613
Replying to @aakashgupta
The slice a stone and get the same results. All fake
1
405
Replying to @aakashgupta
I read the whole thing twice and still couldn’t make sense of what is happening 😢
180
Replying to @aakashgupta
The fly name is "Drosophila"
336
Replying to @aakashgupta
The Matrix.. science fiction leading facts again
501
Replying to @aakashgupta
Fun demo, but no paper yet. FlyWire maps ~139k neurons and 54.5M synapses; Odyssey hasn't said which ones drive the flight.
95