A collaboration of scientists from the Quantum Materials for Energy Efficient Neuromorphic Computing (Q-MEEN-C) consortium, spearheaded by the University of California San Diego, has uncovered a groundbreaking potential in quantum substances. This newly discovered “non-local” behavior could be a key to emulating the human brain’s complex operations, setting the stage for a revolution in energy-saving computing.
The recent advancements in the study of quantum materials have revealed possibilities for creating computers that function like the human brain but with significantly lower energy consumption. The finding of this non-local behavior is a pivotal point, opening the door to a new phase of innovative computing.
While conventional computers excel at tasks like rapid data retrieval and intricate calculations, the human brain’s ability to efficiently and quickly process multifaceted information remains unmatched. For example, the brain’s skill in recognizing faces after just one meeting or differentiating between varied landscapes demonstrates its incredible effectiveness.
The recent discovery in quantum materials has brought us closer to constructing computers that function like the human brain but with a much smaller energy footprint. This discovery has created a path to energy-conserving neuromorphic computing, where devices can replicate the human brain’s complex workings.
According to Professor Alex Frañó, co-leader of Q-MEEN-C, non-local interactions, though typical in the human brain, were previously rare in man-made materials. The new ability to create these interactions provides opportunities to develop more intelligent machines with enhanced learning abilities.
The research, detailed in Nano Letters, builds upon previous efforts to simulate the brain’s components, such as neurons and synapses, using quantum materials. The latest study shows that electrical signals passed between adjacent electrodes can affect other non-adjacent electrodes. This non-locality is of vast importance, signifying an essential step towards brain-like computing.
The journey to this discovery was fraught with obstacles, including pandemic-related lab closures. The team was forced to innovate, using arrays of devices to emulate neurons and synapses, and thus validating the theoretical existence of non-locality in quantum substances.
When labs reopened, the researchers worked with UC San Diego’s Jacobs School of Engineering Associate Professor Duygu Kuzum to translate the simulations into real devices. They manipulated a thin film of nickelate, a special ceramic with intriguing electronic properties. By adding hydrogen ions and an electric signal, the scientists achieved a permanent change in the material’s memory-like structure. This change continued even after the signal stopped, illustrating a major advancement in creating more effective pathways for electric current.
Uniquely, this design concept departs from traditional electrical circuitry. Inspired by the non-local behavior found in the research, the Q-MEEN-C team realized that continuous wire connections were unnecessary. This web-like design, where activity in one part affects the whole structure, leads to a more efficient and economical way to create electrical networks.
Although AI software has achieved great strides in mimicking brain functions, the ultimate potential will only be reached when cutting-edge hardware matches the software’s abilities. This integration of hardware and software is essential for advancing AI. Professor Frañó sees a parallel revolution in hardware that complements current software advancements, and replicating non-local behavior in synthetic materials is a move towards this goal.
As this study continues to advance, the next stage includes developing more complex arrays with added electrodes. This development is guiding us towards a new era in artificial intelligence, where the unification of hardware and software could lead to machines that learn through physical properties, reflecting the intricacy and efficiency of the human brain.
