Cortical Labs has launched the sale of the first computer powered by human or animal neurons. The developers claim that it learns and processes information much faster than classic artificial intelligence models. The company notes that the technology can be used for ethical research into how animal brains work. To conduct research, scientists will be able to purchase either the installation itself with neurons or remote access to it. About this informed New Atlas.
While conventional AI models use silicon semiconductor chips, like most modern computers, to train them, the principle of a biocomputer is to use a natural, efficient system for signal processing and computation, like the brain of an animal or human.
In 2022, Cortical Labs already showed that such a biocomputer could be taught to play Pong, one of the simplest computer games. However, the new model, according to the company, may have much broader capabilities.
The company’s new development, CL1, consists of a box containing neurons derived from human or animal stem cells. The researchers have developed two approaches to growing these neurons: one that allows you to get neurons of one type quite quickly, and the other that allows you to grow different types of neurons and other cells, as in a real brain. To maintain the neurons for six months, the box contains a nutrient-rich liquid that is constantly stirred and oxygenated.
The cells are connected to computers using metal microelectrodes embedded in a glass substrate. Compared to the first design, which used silicon chips, this approach allows for a cleaner signal and avoids the accumulation of charge on the cells, which shortens their lifespan. The box can then output the results of the calculations to other devices using built-in ports.
The CL1 biocomputer is the first step towards creating the smallest complete brain — a project in which a complete brain with the ability to work no worse than a human brain can be created from the smallest number of brain cells. In particular, for the implementation of this project, Cortical Labs encourages researchers from around the world to use their development in areas from drug development to clinical trials to understand the capabilities and limitations of the CL1. Perhaps in the future, such a neuron-based biocomputer will replace the usual one, for example, in controlling robots.
Incidentally, Cortical Labs' success with Pong was replicated by other researchers who used an even simpler system—a hydrogel. Artificial intelligence based on it was also able to play the game and learn in the process.
Thank you for being with us! Monobank for the support of the ElitExpert editorial office.
