Dr. Alex Wissner-Gross on the first multi-behavior fruit fly brain upload — and why mind migration to the cloud is coming in 5-10 years

Summary

Alex Wissner-Gross, physicist, serial founder, and co-founder of Eoghan Systems, is making a straightforward bet that the AI infrastructure buildout now absorbing trillions of dollars in CapEx is almost entirely pointed at artificial minds, and he wants to get biological minds into the cloud alongside them.

Eoghan Systems announced what it is calling the first multi-behavior upload of a fruit fly brain. The project connected the Flywire connectome, a high-resolution map of a specific fruit fly's neural wiring, to a MuJoCo physics-simulated body. The team combined the Flywire dataset, a Nature paper by Eoghan senior scientist Philip Chew demonstrating circuit-level emulation, the NeuroMechFly v2 simulation environment, and coordinated motor actuation work from Erzdahl et al. Eoghan was the first to wire all of these together. Wissner-Gross says the viral response surprised the team, noting the demo began as a side project.

Architecture

The emulation runs on a leaky integrate-and-fire model, a graph of nodes with LIF dynamics firing at each other, closer to a graph neural network than anything in the modern LLM stack. No system prompt is needed to initialize it. Wissner-Gross leaves open the possibility that something analogous to prompt engineering will eventually matter for whole-brain emulation, but there is no equivalent today. The motor outputs use abstract motor movement representations from the Erzdahl paper as the output layer, which is why the demo shows coordinated fly behavior rather than raw spike noise.

Scale gap to humans

The human brain contains roughly 100 to 200 billion cells depending on how you count. The fruit fly connectome is many orders of magnitude smaller. Wissner-Gross draws an explicit parallel to early autonomous vehicle development, describing the fly demo as a level one or level two moment for a field that will need years to define its own fidelity scale.

Fundraising and next targets

Eoghan is beginning fundraising. The next targets are mice and, eventually, humans. Wissner-Gross describes whole-brain emulation as the company's actual goal, with the fly demo representing a small fraction of the broader research program, which also includes scanning connectomes of human and nonhuman animals at scale using expansion microscopy and AI-assisted techniques.

Wissner-Gross on the broader moment

He argues AGI arrived in summer 2020 with GPT-3, and that recursive self-improvement is already underway across the frontier labs. His personal framework is post-singularity, meaning he is making bets on what the world looks like after the transition rather than on who wins during it. Eoghan is one such bet. Another is Physical Superintelligence, a company he describes as using AI to solve open problems in physics. He also co-authored a book with Peter Diamandis arguing that entire academic disciplines will be overrun by superintelligence, and that the strategic question now is which disciplines get targeted first.

His criticism of the current startup ecosystem is direct. Too many new companies are not being ambitious enough and are implicitly betting against the rate of AI capability growth rather than with it. His own portfolio includes startups attempting to grow new landmasses using AI, bets he says would have been unthinkable a few years ago.