Study finds infants outpace AI chatbots in language-learning efficiency
MIT Technology Review spoke with cognitive scientists comparing how babies and large language models acquire language, finding that infants learn to speak proficiently after hearing only 10 to 30 million words, while LLMs require vastly more data to approach similar fluency. Stanford's Michael C. Frank noted that training a model like GPT-2 on the same word count a toddler hears produces a 'nonsense generator,' not coherent speech.
GoKawiil's interpretation of the reporting above, not reported fact.
The comparison highlights a fundamental inefficiency in how AI systems are built, requiring massive datasets and computing resources to mimic what human brains do naturally with minimal input. This gap suggests that scaling data alone may not close the divide between statistical pattern-matching and genuine human-like learning, a challenge companies must address to build more efficient, humanlike AI.
- Babies learn language proficiently after 10-30 million words of exposure.
- LLMs need dramatically more data to reach similar linguistic competence.
- Efficiency, not just scale, remains a core unsolved problem in AI development.
The Language Instinct" by Steven Pinker — If this article's comparison of infant language learning to LLM training hooked you, Pinker's classic dives deep into how the human mind acquires language so effortlessly. It's an accessible, entertaining read that gives real context to why babies still outclass chatbots at learning to speak.
See The Language Instinct" by Steven Pinker on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: futurism.com — Joe Wilkins, 2026-08-25
Published there as: “Even Babies Are Still Way Better at Learning Than AI Models”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.