Young children beat Artificial Intelligence (AI) in learning new language skills according to research co-authored by Lancaster University.
If a human learned language at the same rate as ChatGPT, it would take them 92,000 years. Although computers can handle huge datasets at lightning speed, children are much better than AI when acquiring natural language.
Lancaster Psychology Professors Gert Westermann and Professor Padraic Monaghan from the ESRC International Centre for Language and Communicative Development (LuCiD) co-authored the research in Trends in Cognitive Sciences with Professor Caroline Rowland of the Max Planck Institute for Psycholinguistics and others.
Professor Monaghan said: ""This work highlights just how differently children learn language compared to contemporary AI approaches, and shows the importance of a rich, varied, and social environment for children in the early stages of life."
Professor Gert Westermann said: "Language learning is so much more than hearing language and figuring out what it means. The reason why children learn language with so much apparent ease is because they take an active role in their learning within a rich social and physical world."
Scientists can now observe, in unprecedented detail, how children interact with their caregivers and surroundings, fueled by recent advances in research tools such as head-mounted eye-tracking and AI-powered speech recognition. But despite the rapid growth in data collection methods, theoretical models explaining how this information translates into fluent language have lagged behind.
The new framework addresses this gap. Synthesizing wide-ranging evidence from computational science, linguistics, neuroscience and psychology, researchers suggest that the key to understanding how children learn language so much faster than AI, lies not in how much information they receive, but in how they learn from it.
Unlike machines that learn primarily, and passively, from written text, children acquire language through an active, ever-changing developmental process driven by their growing social, cognitive, and motor skills. Children use all their senses- seeing, hearing, smelling, listening and touching-to make sense of the world and build their language skills. This world provides them with rich, and coordinated signals from multiple senses, giving them diverse and synchronized cues to help them figure out how language works.
Children also actively explore their surroundings, continuously creating new opportunities to learn.
Professor Rowland said: "AI systems process data ... but children really live it. Their learning is embodied, interactive, and deeply embedded in social and sensory contexts. They seek out experiences and dynamically adapt their learning in response - exploring objects with their hands and mouths, crawling towards new and exciting toys, or pointing at objects they find interesting. That's what enables them to master language so quickly."
These insights hold far-reaching implications for research in AI, adult language processing, and even the evolution of human language itself.
"AI researchers could learn a lot from babies. If we want machines to learn language as well as humans, perhaps we need to rethink how we design them-from the ground up."
