Babies Outsmart AI in Fundamental Psychology Duties

Artificial Intelligence is everywhere these days. From self-driving cars to speech recognition, AI is a part of our daily lives. A recent study showed, however, that newborns beat AI in critical psychology tasks.

A new study by New York University published in the journal Cognition suggests infants can exceed artificial intelligence in perceiving motivation behind an individual’s gestures. This study focuses on the importance of improving today’s technology and pinpointing the limitation in AI, as it proved the patent contrast between cognition and computation.

"Adults and even infants can easily make reliable inferences about what drives other people’s actions," said an assistant professor in New York University’s Department of Psychology and the paper's senior author, Moira Dillon, Ph.D. "Current AI finds these inferences challenging to make."

She continues, "The novel idea of putting infants and AI head-to-head on the same tasks is allowing researchers to better describe infants’ intuitive knowledge about other people and suggest ways of integrating that knowledge into AI."

Infants are captivated by others, as proved by how they stare at and perceive humans. They also communicate with others and can understand the rudimentary feelings of individuals. Babies develop human social intelligence through their abilities by developing goals and having specific preferences.

The study delved into the difference between an infant and AI by conducting a study with 84 11-month-year-old babies. The team contrasted infants to AIs and observed their reactions to the 'state-of-the-art learning-driven neural-network model.' The team used the "Baby Intuitions Benchmark" (BIB), six tasks that probe commonsense psychology.

BIB was created to test infant and machine intelligence, permitting a comparison of newborn and machine performance and, most importantly, supplying an observed foundation for developing humanistic AI. On Zoom, infants observed videos with simple, animated shapes rolling around the screen. Human behavior and decision-making were simulated by retrieving objects displayed on the screen and other movements.

Likewise, the team constructed, trained, and tested learning-driven neural network models, or AI machines, that assist computers in identifying patterns and imitating human intelligence. The team found newborns could discern human-like motives in simple actions and animated shapes. They were to identify the retrieval of identical objects on the screen despite the constant environmental changes. Newborns stared at moving objects longer, indicating recognition.

On the other hand, AI tools failed to exhibit any evidence of recognition. This capacity to ponder actions and information appears to be distinctive to humans as of now. It lets us connect with other people and work together.

Dillon concluded: "A human infant’s foundational knowledge is limited, abstract, and reflects our evolutionary inheritance, yet it can accommodate any context or culture in which that infant might live and learn."