A Fruit Fly's Brain Is Now Generating Magazine Story Ideas, and It's Not as Crazy as It Sounds
Researchers mapped every neuron in a real fly's brain and open-sourced it. Now people are using that map to pitch articles, trade stocks, and play video games. Wired AI's Will Knight reports on where this is actually heading.
Key points
- Google and academic researchers published a connectome (a complete wiring diagram of every neuron and synapse) for a male fruit fly in early September 2025.
- The map covers 166,000 neurons and 125 million synapses and is free for anyone to use.
- Wired AI writer Will Knight used it to build PitchFly, a tool that reads popular headlines and generates new story ideas.
- A Coinbase software engineer built StonkFly, which uses the same fly brain to trade stocks. It is losing money but outperforming expectations.
- Neuroscientists say mapping small animal brains can reveal insights into how human intelligence and neurological damage work.
A fruit fly has 165,112 neurons. You have roughly 86 billion. A digital copy of that tiny fly brain is now pitching magazine articles, trading stocks, and trying (badly) to park virtual cars.
This all starts with a connectome, a complete map of every neuron and every connection between neurons in a brain. Think of it as a wiring diagram for the entire electrical system of a living creature. Researchers from Google and several academic institutions published a connectome for a male fruit fly in early September 2025, capturing 166,000 neurons and 125 million synapses, the chemical bridges that carry signals between nerve cells. They made it free to use.
What has anyone actually built with it?
Quite a lot, very fast. Will Knight at Wired AI used the connectome to build PitchFly: he fed hundreds of the site's most-read headlines into the fly brain and had it remix the patterns into new story ideas. A favourite output was "Everyone Wants Cooking. Nobody Has Solved Donald Trump."
PitchFly does not understand words. It cannot. The fly brain simply spots patterns in shapes and sequences, then recombines them. It is closer to a kaleidoscope than a journalist.
| Project | Creator | What it does | Result |
|---|---|---|---|
| PitchFly | Wired AI writer | Generates magazine story ideas | Entertaining nonsense |
| StonkFly | Alex Wormuth, Coinbase | Trades stocks | Losing money, but surprisingly close |
| Beat Saber trainer | X user Lyra Bubbles | Plays a VR rhythm game | Demo'd publicly |
| Doom / Minecraft / Pong | Various developers | Plays classic video games | Mixed results |
| Parallel parking | Unknown | Drives virtual cars | Flies are terrible at it |
Alex Wormuth, the Coinbase software engineer behind StonkFly, told Knight in a direct message that the project felt philosophically interesting beyond the laughs. "It sparks questions about ethics of this and whether there is any consciousness in the replication of a fly's brain," Wormuth said. StonkFly is losing money, for the record. It's doing better than you'd expect from an insect.
Why does any of this matter beyond a novelty?
Two serious things are happening at once here.
For neuroscientists: mapping small brains is a practical way to study how intelligence works at a biological level. Researchers can simulate what happens when specific neurons are damaged, which could eventually inform how we treat neurological conditions in humans. The full human brain, at 86 billion neurons, is still far beyond current mapping technology. Google Research has featured in three of our stories in the past 90 days, and the pattern is consistent: open-source releases from that lab tend to generate downstream projects faster than anyone anticipates.
For anyone watching AI: the ease with which dozens of developers built working projects on top of a published connectome in weeks is a real signal. You don't always need a giant, expensive model to do something useful. A tiny brain, pointed at the right task, can be enough.
That's the honest takeaway. Not that fruit flies are secretly smart. The tools for building specialised AI models have become cheap and accessible enough that a single engineer at home can train a digital insect to trade real stocks. That shift matters more than the novelty.
Before you follow Wormuth's lead: StonkFly is losing money. Start with the novelty, not your savings account.
Common questions
Does this mean AI is now copying real brains?
Not quite. The fruit fly connectome is a map of a biological brain, not a model built to imitate one. Using that map to power software is a novel approach, but it is a long way from replicating human cognition.
Could a fruit fly brain replace an AI chatbot for some tasks?
For narrow, pattern-based tasks, yes, and far more cheaply than a large language model. For anything requiring understanding of meaning or context, no. The fly has no idea what words mean; it shuffles patterns it has seen before.
