FlyWire finished the wiring diagram. The function did not follow.

On October 2, 2024, Nature published a nine-paper package from the FlyWire consortium. The team, led by Sebastian Seung and Mala Murthy at Princeton with more than 200 researchers across 127 institutions, had completed the first whole-brain connectome of an adult Drosophila melanogaster. 139,255 neurons. 50 million synaptic connections. Five years of AI-assisted segmentation and human proofreading, contributed by online gamers as well as researchers, equivalent to 33 person-years of validation work. Without AI, the project would have taken 50,000 person-years.

The achievement is real. The interpretive lesson is harder. A complete wiring diagram does not, on its own, yield a model of how the fly thinks, navigates, or remembers. The connectome is the static graph. The function is the dynamics over that graph, and the dynamics depend on neurotransmitter type, firing patterns, plasticity, neuromodulation, and embedded prior structure that the wiring diagram does not specify.

The optic-lobe analyses that came with the nine-paper package made this explicit. Connectivity alone produced theoretical predictions; matching those predictions to behavior still required experiments. The connectome is necessary but not sufficient.

For AI, the lesson is the inverse of the popular framing. The popular framing is that whole-brain connectomes will give AI builders the architectural specification they need. The lesson FlyWire actually demonstrates is that even when you have the wiring of an entire brain, you still cannot derive the algorithm from the structure. The connectome enables hypothesis generation. The hypotheses still have to be tested. Anyone hoping that mouse or human connectomes will hand AI a blueprint is misreading what FlyWire just established.