🧠 This ultra-thin brain implant enables efficient connection of the brain to an AI

How can we interact with computers using only our mind? Existing brain-computer interfaces are often limited by their size and invasiveness, requiring major surgeries and offering low data rates.

A recent innovation, developed by several universities, proposes a surprising solution: a brain implant as thin as a hair, capable of communicating wirelessly at extremely high speeds. This device, called BISC, could transform how we treat neurological disorders and interact with technology, by creating a direct and powerful link between the brain and machines.

Developed by teams from Columbia University, NewYork-Presbyterian Hospital, Stanford University, and the University of Pennsylvania, the BISC is based on a unique silicon chip. This ultra-thin chip can be inserted into the space between the brain and the skull, thereby reducing surgical risks. With over 65,000 electrodes, it records and stimulates brain activity with unprecedented precision, allowing artificial intelligence algorithms to decode intentions and perceptions (explanation at the end of the article). The researchers published their results in Nature Electronics, detailing how this approach pushes the boundaries of traditional brain-computer interfaces.


The technology behind the BISC integrates all electronic components onto a single CMOS chip, reducing its volume to less than one-thousandth that of conventional implants. This remarkable miniaturization, enabled by processes from the semiconductor industry, facilitates insertion through a small incision, unlike older devices that used bulky canisters. Consequently, this technical advance minimizes invasiveness and improves the stability of neural recordings over time, according to the engineers' explanations. It also represents a notable step in the evolution of electronic chips for medical implants.

Furthermore, with a data rate reaching 100 Mbps via an ultra-wideband radio link, the BISC transmits brain information to advanced machine learning tools. These tools can interpret elaborate neural activity patterns, paving the way for promising medical applications. For example, it could help manage drug-resistant epilepsy or restore motor functions after a spinal cord injury. Preclinical trials have already shown encouraging results, and preliminary human studies are underway to validate these benefits.

To accelerate its adoption, the researchers founded Kampto Neurotech, a spin-off company developing commercial versions of the chip. In the long term, the BISC could enable smooth interaction between the brain and artificial intelligence systems, going beyond disease treatment.

Decoding neural signals by artificial intelligence

The artificial intelligence algorithms used with brain-computer interfaces analyze data from the electrodes to interpret brain activity. They work by learning to recognize specific patterns in the electrical signals generated by neurons, which correspond to thoughts, intentions, or perceptions. To illustrate, when you imagine moving your hand, certain areas of the motor cortex activate, and the AI can associate this pattern with the corresponding action.

This approach relies on deep learning techniques, where artificial neural networks are trained on vast sets of neural data. Researchers collect this data by recording brain activity while subjects perform specific tasks, such as viewing images or attempting to speak. The AI then learns to predict mental states from these recordings, improving its accuracy over time and enabling real-time applications.

In the medical field, this decoding can help restore lost functions, such as speech or mobility, by translating intentions into commands for prostheses or stimulators. Recent advances, supported by devices like the BISC, increase the speed and reliability of these processes, opening the door to personalized therapies. However, obstacles remain, such as individual variability in brain signals, requiring continuous adaptation of AI models.