In a new study, researchers conducted a series of experiments with 11-month-old infants, and compared their responses to those of state-of-the-art learning-driven neural-network models.
Infants are fascinated by other people and observe their actions. They attribute goals to others and expect others to pursue goals rationally and efficiently. Conversely, “commonsense AI” — driven by machine-learning algorithms — predicts specific actions directly.
Infant understanding vs. algorithms
In the research, infants watched a series of Zoom videos with simple animated shapes moving around the screen — similar to a video game. The shapes’ actions simulated human behavior and decision-making. The researchers then compared their responses to those yielded by state-of-the-art learning-driven neural-network models with the same videos.
Results showed that infants recognize human-like motivations, even in the simplified actions of animated shapes; while the algorithm-based models showed no such evidence of understanding the motivations underlying the actions. Perhaps like the notorious Bing chatbot saying "I love you!"
Citation: Stojnić, G., Gandhi, K., Yasuda, S., Lake, B. M., & Dillon, M. R. (2023). Commonsense psychology in human infants and machines. Cognition, 235, 105406. https://www.sciencedirect.com/science/article/pii/S0010027723000409