The so-called “neuromorphic computing” is the process by which computer chips are designed and manufactured with the aim of reproducing the structure and operation of the human brain. In this way, using artificial neurons and synapses, computers simulate the way in which people manage information, allowing us to solve problems that previously seemed unsolvable.
Thanks to this advancement, machines can recognize patterns better than in the past, making it easier for them to make faster decisions. The inspiration for this system, as with many others, is intellect, as Accenture Labs associate director Andreea Danielescu points out. Its application is scarce, incipient. For now, it does not go beyond universities, administrations and large corporations such as IBM or Intel.
However, analysts and engineers agree that few technologies are as promising as neuromorphic computing. Their intervention will improve the functioning of artificial intelligence, for example, in cases such as autonomous vehicles, where speed and efficiency are imperative. And the key to this revolution is in the cortex of the brain.
This is where higher cognitive missions are located, such as sensory perception, motor commands, spatial reasoning and even language. The layered structure and intricate connectivity in the neocortex are fundamental in processing complex information. This is one of the bases of human thought and transferring it to computers has not been easy.
Neuromorphic computers orchestrate spiking neural networks, which originate when their active neurons, which contain biological-like data, are linked through artificial synaptic devices that transfer electrical signals. This logic leaves behind the binary (zeros and ones) and sequential approach. The new computers perform as they should, without material jams at critical moments.
Popular tools like ChatGPT need it to progress. Quantum computing is especially good at tackling complicated challenges, such as those posed by cryptography or molecular simulation. The equipment requires lower temperatures and more energy than the neuromorphic formula, which is capable of learning in real time. Furthermore, experts conclude that logistically it is preferable and cheaper than the quantum version.
