Why AI Still Struggles Against the Human Brain
News Desk
Islamabad: Despite rapid advances in artificial intelligence, the human brain continues to outperform computers when it comes to learning across different tasks and transferring skills from one challenge to another, a new study suggests.
Researchers from Princeton University explored how this flexibility works by studying rhesus macaques (Macaca mulatta), whose brain structure and function closely resemble those of humans. Instead of testing people directly, the team observed how the monkeys’ brains responded while they completed simple visual tasks.
The animals were asked to identify shapes and colours displayed on a screen and respond by directing their gaze in specific directions. During the exercises, researchers monitored brain activity to identify shared patterns across different tasks.
The scans revealed that the monkeys’ brains relied on distinct yet reusable groups of neurons — described by researchers as “cognitive Legos.” These neural building blocks can be rearranged and reused to perform new tasks, allowing the brain to adapt quickly in ways current AI systems cannot.
“State-of-the-art AI models can match or even exceed human performance on single tasks,” said neuroscientist Tim Buschman of Princeton University. “But they struggle to learn and perform many different tasks.”
According to the study, the brain’s strength lies in its ability to recombine existing cognitive components rather than learning every task from scratch. When certain neural blocks are not needed, their activity is reduced, enabling the brain to focus more efficiently on the task at hand.
Buschman compared these cognitive blocks to functions in a computer program, where one set of neurons may process colour while another translates that information into an action. This step-by-step organization allows the brain to solve complex problems using familiar components.
The findings help explain how humans and animals can adapt to unfamiliar situations by applying prior knowledge — a capability that remains a major challenge for artificial intelligence in its current form. Input from Duniya news.
Comments are closed.