Scientists at Cold Spring Harbor Laboratory (CSHL) have created an AI algorithm inspired by how efficiently our genome stores information. This breakthrough could explain how the brain manages to control complex tasks with limited genetic instructions.
Many animals, like spiders spinning webs or whales swimming, show amazing skills right after birth. But how do they know what to do without learning first? The brain is essential, but the human genome only has space for a small part of the information needed for such abilities. This has puzzled scientists for years.
CSHL Professors Anthony Zador and Alexei Koulakov believe the genome’s limits might be the secret to our intelligence. Zador suggests, “Maybe having less information in the genome forces us to adapt and learn faster.”
Using AI to Copy Nature’s Efficiency
To test this idea, Zador and Koulakov, along with researchers Divyansha Lachi and Sergey Shuvaev, developed an AI algorithm that compresses large amounts of data. This is similar to how our genome might pack complex instructions into a small space.
When compared to advanced AI models, their new algorithm, without any training, performed nearly as well in tasks like image recognition and even playing video games like Space Invaders. It seemed to understand the task almost naturally.
Although AI isn’t yet close to matching human brain abilities, this algorithm shows promise. Koulakov explains, “The human brain can hold about 280 terabytes of information, but our genome only stores a tiny fraction of that—equivalent to one hour of video.”
Even though we aren’t there yet, this new AI method could lead to more efficient technology, such as running large AI models on small devices like smartphones. Lead researcher Shuvaev says, “This could make AI models faster and more accessible in everyday tech.”
After billions of years of evolution, scientists are now using AI to uncover the secrets of how our brains work—and it could change the future of technology.
Source: scitechdaily
