The current machine learning utilizes software in order to process and store information. The energy usage of this process is extremely inefficient, however, leading to questions of how this process could be sped up.
Recently, researchers physicists at Radboud University were able to speed up this process through the usage of hardware instead of software. Through their research, they discovered that they could create a network of cobalt atoms on black phosphorus, creating a material which processes information much faster and much more efficiently than machine learning. Similar to the human brain, the material was able to adapt, changing the state of the atoms within the network depending on the input.
It will be interesting to see where this new material will be incorporated. Although there is still much research to be done, it already seems like the development of better hardware could drastically increase the capabilities of machine learning, and could also make the process much more energy efficient. I think that the material could also be used in quantum computing. Maybe new materials like this could be used to create more qubits or help make our current qubits achieve higher fidelity and gate speeds, speeding up the creation of a universal quantum computer.
“An atomic Boltzmann machine capable of self-adaption” by Brian Kiraly, Elze J. Knol, Werner M. J. van Weerdenburg, Hilbert J. Kappen and Alexander A. Khajetoorians, 1 February 2021, Nature Nanotechnology.
O’Neill, Mike. “The First Steps Toward a Quantum Brain: An Intelligent Material That Learns by Physically Changing Itself.” SciTechDaily, 7 Feb. 2021, scitechdaily.com/the-first-steps-toward-a-quantum-brain-an-intelligent-material-that-learns-by-physically-changing-itself/.