h+ Magazine

Yet another tiny computer? The press release from Michigan Tech asserts that it is “the first time a brain-like ‘evolutionary circuit’ has been realized.” This new molecular computer uses an organic molecular layer and can evolve to solve complex problems, similar to neurons. Like the human brain — and unlike any existing computer — the tiny molecular computer heals itself if there is a defect. Anirban Bandyopadhyay, from the Japanese National Institute for Materials Science, explains: “No existing man-made computer has this property, but our brain does. If a neuron dies, another neuron takes over its function.”

The international research team from Japan and Michigan Tech compared the evolving patterns of their molecular computer to neural activity in the brain as seen by a Functional MRI. With the ability to image the entire volume of the brain, fMRI can isolate many simultaneous and coordinated neural events, as shown in this video of a 3D fMRI of the human brain:

Interestingly, the evolving patterns generated on the molecular layer — when viewed with a scanning tunneling microscope — bear an uncanny resemblance to fMRI images of various events in the human brain. The researchers next used the electrically charged tip of the scanning tunneling microscope to individually set molecules in the top layer to a desired state, essentially writing data into the system.

Molecular computing is most frequently associated with DNA computing: custom-synthesized DNA strands are combined with enzymes to create output strands. DNA computing is limited by its ability to return only yes-or-no answers (0, 1) like a binary computer — one strand is longer than another. This results in static circuits that are unable to evolve. In contrast, the electrical circuits of our brains are dynamic, ever-changing and evolving networks of neurons that process events in parallel. While conventional computers are typically built using two-state (0, 1) transistors, the molecular layer is built using a hexagonal molecule called DDQ, made of nitrogen, oxygen, chlorine and carbon, that self-assembles in two layers on a gold substrate and can switch among four conducting states — 0, 1, 2 and 3. “The neat part is, approximately 300 molecules talk with each other at a time during information processing,” says Physicist Ranjit Pati of Michigan Tech. “We have mimicked how neurons behave in the brain.”

The recently-published research paper on the organic molecular layer in Nature Physics explains how modern digital supercomputers operate sequentially at speeds of up to ten trillion instructions per second, and while individual neurons in the human brain fire at only approximately 1000 times per second, they do so collectively and in parallel. Unlike the brain, however, the simultaneous, parallel processing of the DDQs used in the molecular monolayer can produce solutions to problems for which algorithms on computers are unknown, like predictions of natural calamities and outbreaks of disease. The supplementary materials include videos showing the creation of the organic layers, the construction of circuits, and how to replicate heat diffusion and cancer cell evolution.

Quantum computing is yet another form of molecular computing that allows for multiple simultaneous computations. It takes advantage of quantum effects to perform the computations and is dependent upon supercooled atoms locked in entangled states with one another. As the number of computational elements (qubits) increases, it becomes progressively more difficult to insulate a quantum computer from matter on the outside. Qubits are different than traditional binary bits. Instead of being yes-or-no, qubits can hold a value of both 0 and 1, as well as all values between 0 and 1. Qubits can be made of atoms, ions, photons or electrons, and can easily become destabilized and lose their values.

Does the organic molecular layer have more AI potential than quantum computing? The Michigan Tech announcement is almost unbelievable — a 4-state organic molecular layer that demonstrates brain-like capabilities such as massively parallel processing, the ability to evolve and self-heal, all while acting like a human brain as seen on a fMRI brain scan. The organic molecular layer might work as a substrate for advanced AI, assuming a way can be developed to write data faster than using a scanning tunneling microscope.