The development of a biological computer with millions of human brain cells is proposed by dozens of scientists in the United States, counting that in the future such a device will be able to outperform conventional computers based on silicon microchips in performance, consuming much less energy.

An international research team led by scientists at Johns Hopkins University in Baltimore published last Tuesday in the journal Frontiers in Science a detailed blueprint for what they call “organ-like intelligence.”

The biological computer will include arrays of brain organoids, i.e. tiny neural structures made from human stem cells in three dimensions, that will be connected to sensors and outputs and trained using machine learning.

The goal is to develop a highly efficient computing system that can solve problems that conventional computers cannot. At the same time, the researchers hope to advance the development of neuroscience and other areas of medical research. The project moves in the same scientific field as research into more advanced quantum computers, “but raises ethical questions related to consciousness” of brain organelle systems, the Financial Times comments.

Brain without the limits of the human body

The scientists envision “an intelligent dynamic system that will be based on synthetic biology, but not limited to the many functions that the brain must perform in the body,” explains Johns Hopkins University professor Thomas Hartung, who assembled a team of 40 scientists. for the development of this technology.

The scientists signed the “Baltimore Declaration” calling for further research into “the potential of organoid cell cultures to improve our understanding of the brain and harness new forms of biocomputing” while recognizing and addressing emerging ethical issues. Hartung acknowledged that developing artificial intelligence on a commercial scale would likely take decades.

Ethical dilemmas

In addition to scientific goals, there are ethical dilemmas about creating a “mind on a plate” that can learn, remember and interact with the environment – and can develop consciousness even in a rudimentary form. Hartung replies that ethical issues will be constantly evaluated by groups of scientists as well as society.

Madeleine Lancaster, a brain organ researcher at the Cambridge Molecular Biology Laboratory, is not part of the team and has not disclosed the purpose of the project. “It’s really science fiction, and while it’s interesting, it lacks a scientific basis,” he said. “There are huge hurdles to overcome in order to do what they offer,” he comments. Karl Friston, a non-AI professor of neuroscience at University College London, is more positive. “This is definitely an idea worth exploring,” he said. “Many small steps must be taken, but their direction travel it could be revolutionary.”

A necessary step was the development of individual organelles with the invention of a better way to feed them with nutrients in laboratory dishes. These tiny neural structures should grow from about 50,000 cells today to about 10 million.to function at a level that scientists would call “organ-like intelligence.”

Researchers are also developing technologies to connect organelles in order to communicate with them, send information and decipher their “thoughts”. Hartung’s lab is experimenting with an interface, “a flexible sheath tightly covered with tiny electrodes that can both receive signals from an organelle and transmit them to it.”

Energy saving

One of the reasons for the shift to bioinformatics is that the brain processes and stores information very efficiently. Launched last year, the world’s most powerful Frontier supercomputer at Oak Ridge National Laboratory in the United States is equivalent to the human brain in terms of processing power – an exaflop, or trillion billion operations per second – but consumes a million times more energy. from the midbrain.