🧠 Here is the first living 3d neural network for AI

A team from Princeton has designed a three-dimensional device that merges living neurons with electronics. This system can be programmed to recognize electrical patterns. Unlike previous attempts, which used flat 2D cultures or cell clusters, the innovation here lies in direct interaction with neuronal cells from inside the network.

To achieve this, the researchers used fabrication methods to create a 3D mesh of microscopic metal wires and electrodes, held together by a very thin epoxy coating. This flexible structure allows tens of thousands of neurons to grow around it, forming a dense three-dimensional network capable of performing calculations. The study was published in Nature Electronics.


This integrated format allows recording and stimulating the electrical activity of neurons with far greater precision than previous systems. For over six months, the team observed the network's evolution, tested methods to strengthen or weaken connections between neurons, and then trained an algorithm to identify patterns in electrical impulses. In two separate experiments, the system correctly distinguished different spatial and temporal patterns.

One of the lead authors, Tian-Ming Fu, explains that the real bottleneck for artificial intelligence in the near future is energy. Our brain consumes about one millionth of the power of current AI systems for similar tasks. This device could therefore help overcome this energy challenge while replicating certain capabilities of the brain.

Kumar Mritunjay, first author of the study, indicates that these 3D biological neural networks do not just reveal the brain's computation mechanisms; they can also help understand and potentially treat neurological diseases.