Brains and artificial intelligence (A.I.). Both are able to do a wide range of tasks from solving a simple math problem to understanding someone’s emotions. What makes A.I. The difference is how much energy is consumed for them “to think”. A human’s brain only uses less power than a light bulb, while A.I. sucks tons of energy to do a simple task.
Now, scientists have made a pocket-sized A.I. that can help scientists have a better idea of how our brains work with the little energy they use. This bot they created is made to copy how the brain’s visual works and used to need 60 million variables, but the team of scientists were able to shrink it down to only 10,000 variables.
Since this A.I. brain seems to work like any other brain, scientists think they can use it to see how Alzheimer’s happens and what goes wrong and overall, how the human brain works.
The idea that gave scientists this idea was the difficulty to figure out how the brain can visually recognize different objects like a pencil or a watch. There’s never been any distinct way to figure out more than the basics. That’s where the A.I. brains come in, where it could possibly explain it to the scientists, but this is a difficult task to get to. Ben Cowley, the author of the study and assistant professor says, “We’re very impoverished in our understanding of how these AI systems work,” Ben Cowley, the author of the study and assistant professor says, “much like our own brain.”
So instead, Cowley created an AI model that would be easier to understand. To do this, he mimicked only one part of the visual system. Other AI can do this, but Cowley’s plan was to make his model more efficient. “We want to take these big clunky models and try to compress it down into a much smaller, compact form,” he says.
They used data from macaque monkeys and took away data that was redundant and created the model. This model will be very useful as time goes on to see how the brain works and neutrons.