In the last year or so, AI models like ChatGPT and DALL-E have made it possible to generate massive amounts of (seemingly) human, high-quality creative content from a simple set of instructions.
See also: Is the human brain similar to a quantum computer?
Despite being incredibly powerful – outperforming humans at big data pattern recognition tasks – today’s AI systems are not intelligent in the same way that we are. AI are not structured in the same way as our brains, and they don’t learn in the same way.
AIs also consume huge amounts of energy and resources during training (compared to our three or more meals each day). Compared to us, their ability to adapt and operate in dynamic, unpredictable, and noisy situations is poor, and they lack the memory capabilities of humans.
The research focuses on non-biological systems that most closely resemble human brains. In a new study published in Science Advances, it was discovered that self-organized networks of tiny silver wires appear to learn and remember in the same way as our brain's thinking circuitry.
The brain's imitation
This is part of a research project called neuromorphics, which aims to replicate the structure and functionality of biological neurons and synapses in non-biological systems.
The 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 a millimeter wide, the width of a human hair. They are made of a highly conductive metal, such as silver, usually coated with 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 through the insulating layer to a neighboring nanowire (like neurotransmitters at synapses). As a result, we observe synaptic-like electrical signaling in nanowire networks.
Proposal:AI: Needs sleep like the human brain
Learning and memory
This nanowire system is used to investigate the topic of human intelligence. Two important features of mental function are crucial to our research: learning and memory.
Our findings show that synaptic connections in nanowire networks can be selectively strengthened (or weakened). This is analogous to “supervised learning” in the brain.
The output of the synapses is compared to a desired outcome in this process. The synapses are then either strengthened (if their output is close to the desired outcome) or pruned (if their output is not close to the desired outcome).
This finding was based on demonstrating how we can increase the amount of reinforcement by “rewarding” or “punishing” the network. This method is based on “reinforcement learning” in the brain.
Artificial intelligence
Human intelligence is very likely still a long way from being copied.
Regardless, research into neuromorphic nanowire networks shows that it is possible to implement features essential for intelligence – such as learning and memory – in a non-biological, natural material.
Nanowire networks are different from the networks used in AI. However, they could lead to "artificial intelligence"
Perhaps a neuromorphic nanowire network could one day learn to have more human-like conversations than ChatGPT and remember them.
Read also: The AI development halt letter and the path of GPT-5
source of information: sciencealert.com
