Researchers have been skeptical that recent AI advances can tell us much about human learning and development. To address this, a team training an AI model, not on massive data, but on the input that a single child receives. Their findings showed that the model could learn a substantial number of words and concepts using limited slices of what the child experienced.