Advertisement

Scientists combine light, superconductors to power large-scale AI

According to at least one researcher at the National Institute of Standards and Technology, superconductors are superior to semiconductors as a medium for integrating optical and electronic components. Photo by Jeffrey Michael Shainline
According to at least one researcher at the National Institute of Standards and Technology, superconductors are superior to semiconductors as a medium for integrating optical and electronic components. Photo by Jeffrey Michael Shainline

April 20 (UPI) -- To build the next generation of artificial intelligence systems, at least one scientist wants to combine photonic components with superconducting electronics.

The development of AI models capable of approximating human cognition have traditionally relied on semiconductors, but the process of combining optical and electronic components on silicon chips is beset by both physical and practical problems.

Advertisement

In a new paper, published Tuesday in the journal Applied Physics Letters, Jeffrey Shainline, engineer and researcher at the National Institute of Standards and Technology in Colorado, calls on scientists in the field of AI to ditch semiconductors in favor of superconductors.

"We argue that by operating at low temperature and using superconducting electronic circuits, single-photon detectors and silicon light sources, we will open a path toward rich computational functionality and scalable fabrication," Shainline said in a press release.

RELATED AI system allows smart speakers to wirelessly monitor heart rhythms

In the paper, Shainline analyzes the benefits of combining optics and electronics, as well as the advantages of superconductors as a medium for optoelectronic integration.

The study provides a roadmap for integrating photonic components with superconducting electronics.

"What surprised me most was that optoelectronic integration may be much easier when working at low temperatures and using superconductors than when working at room temperatures and using semiconductors," said Shainline.

Advertisement
RELATED Scientists explore whether robots can make good friends

One of the advantages of working with superconductors is that they can detect single particles of light, or photons.

Typically, semiconductors can detect optical inputs no smaller than 1,000 photons. As a result, superconductors can operate at much lower energy levels.

Because superconductors require frigid temperatures, Shainline's proposal isn't intended for use in everyday devices like cellphones. Instead, Shainline envisions the use of superconducting hardware in the construction of high powered computing systems.

RELATED Machine-learning program imagines a protein's many possible structures

In future experiments, Shainline said he hopes to investigate the ability of superconducting hardware to replicate the functionality of synapses and neurons in the human brain.

According to Shainline, superconductors could even be used to develop hybrid neural-quantum systems.

Latest Headlines