In the past year or so, generative AI models such as ChatGPT and DALL-E have made it possible to produce large quantities of apparently human-like, high-quality creative content from a simple series of signals.
Although highly capable – far ahead of humans in particular big-data pattern recognition tasks – current AI systems are not as intelligent as we are. AI systems are not structured like our brains and do not learn in the same way.
Also uses AI systems Huge Amount of energy and resources for training (compared to our three or more meals). Their ability to adapt and function in dynamic, unpredictable and noisy environments is poorer than ours, and they lack human-like memory capabilities.
Our research explores non-biological systems that are more like the human brain. In a new study published in Science AdvancesWe found that self-organizing networks of tiny silver wires appear to learn and remember just like the thinking hardware in our heads.
Simulating the brain
Our work is part of a field of research called neuromorphics, which aims to mimic the structure and functionality of biological neurons and synapses in non-biological systems.
Our research focuses on a system that uses a network of “nanowires” to mimic neurons and synapses in the brain.
These nanowires are tiny wires about one-thousandth the width of a human hair. They are made of a highly conductive metal such as silver, usually coated in an insulating material such as plastic.
The nanowires self-assemble to form a network structure similar to a biological neural network. Like neurons, which have an insulating membrane, each metal nanowire is coated with a thin insulating layer.
When we stimulate the nanowires with electrical signals, ions migrate across the insulating layer and into neighboring nanowires (like neurotransmitters across a neurotransmitter). As a result, we observe synapse-like electrical signals in nanowire networks.
Learning and memory
Our new work uses this nanowire system to explore the question of human-like intelligence. At the center of our investigation are two traits that are indicative of higher-order cognitive functioning: learning and memory.
Our study shows that we can selectively strengthen (and weaken) synaptic pathways in nanowire networks. This is similar to “supervised learning” in the brain.
In this process, the output of the synapse is compared with the desired outcome. Synapses are then strengthened (if their output is close to the desired outcome) or pruned (if their output is not close to the desired outcome).
We expanded on this result by showing that we can increase the amount of reinforcement by “rewarding” or “punishing” the network. This process is driven by “reinforcement learning” in the brain.
frameborder=”0″ allowed=”accelerometer; autoplay; clipboard-writing; encrypted-media; gyroscope; picture-in-picture; web-share” allowfullscreen>
We also implemented a version of the test called the “n-back task” that is used to measure working memory in humans. It involves presenting a series of stimuli and comparing each new entry with the one that occurred several steps (n) earlier.
The network “remembers” previous cues for at least seven steps. Curiously, seven is often considered the average number of items humans can hold in working memory at one time.
When we used reinforcement learning, we saw dramatic improvements in the network’s memory performance.
In our nanowire networks, we found that the structure of a synaptic pathway depends on how that synapse has been activated in the past. This is also the case for synapses in the brain, where neuroscientists call it “metaplasticity”.
artificial intelligence
Human intelligence is still a long way from being replicable.
However, our research on neuromorphic nanowire networks shows that it is possible to implement features necessary for intelligence – such as learning and memory – in non-biological, physical hardware.
Nanowire networks are different from artificial neural networks used in AI. However, they can lead to so-called “artificial intelligence”.
Perhaps neuromorphic nanowire networks can one day learn to communicate and remember more human-like interactions than ChatGPT.
Alon Loeffler, PhD Researcher, University of Sydney and Zdenka Kuncik, Professor of Physics, University of Sydney
This article is republished from The Conversation under a Creative Commons license. Read the original article.
#Networks #silver #nanowires #similar #human #brain #learning #memory
