Artificial intelligence often uses a concept called neural networking, a process that involves simulating the synapses of the human brain in computer software. Recent developments have been made in analog neural computing such that neurons are being simulated at the hardware level, not the software level.


The resulting efficiency may someday reach the femtoJoule (fJ) (1×10^-15 J) level, like human brains human brains use between 1 and 100 fJ per neuron fired. It is worth noting that 1 J is roughly equal to lifting a medium sized tomato about 3ft into the air.


This new technology was named by Stanford University, the “electrochemical neuromorphic organic device (ENODe)”, after the method of the machine’s operation and development. If all goes well, the neuron-like processing units may become mass produced and be placed into chips and human-machine interfaces on a society-wide scale.


How does neuromorphic processing work?


Neuromorphic processing works very similarly to the human brain in the case of this new synaptic machine. The machine emulates synapses by integrating conductive silver particle clusters into a non conductive film. The film is placed over electrodes which, when heated, cause the film to expand and touch its neighboring film structure. When electrical input is halted, the films retract.


By using similar neuron mimicking hardware, hardware developers have been able to stuff as many as 5.3-billion transistors, 256 million programmable, synthetic synapses, and 4096 neurosynaptic cores. The resulting processor can handle as many as 500 programmable, computable states, compared to the 2 states of traditional computing (1 or 0).


What effects will it have in artificial intelligence and computing?


By increasing the effectiveness of memory pathways, reducing electricity usage and heat, and further reducing the size of processors, neuromorphic devices are paving the way for considerably more space and energy efficient, smarter artificial intelligences. Despite current advancements, it is estimated that modern neuromorphic computers are nearly 10,000 times more energy inefficient than the human brain.


ENODe technology is poised to upset the traditional method of computing advancement, allowing for a more natural and efficient design. Traditional computing components are literally millions to billions of times more inefficient than the human brain, but ENODe technology advancement may change that. ENODe computing may not be best applied to every computing field, however.


How will neuromorphic processing change the way humans and machines interact?


Neuromorphic processors are paving the way for significantly improved brain-computer interfaces, allowing humans to carry out tasks such as teaching AIs and operating their daily lives with increased efficiency.


Affected tasks may include driving, machinery operation, gaming, work, lifestyle planning, and eventually most of life activities. ENODe technology may allow for truly immersive virtual reality spaces based on human-machine interaction at the electrical level.



ENODe computing is one of the most disruptive computing breakthroughs of 2017 so far, alongside Li-RAM which can be viewed in our article Computer Memory Access at the Speed of Light. By increasing the efficiency of modern computing to the million or billions of times scale, power consumption can be greatly decreased, reducing the effects of fuel consumption.


Other areas where the technology can save on power include driving activities, production, and supply chain operations.



As the needs of society expand, so do the abilities of the technologies which fuel its advances. With every breakthrough in terms of efficiency, others can occur in all other aspects of society operation, ranging from healthcare to transportation and warfare.