The AI program learned its first words by observing objects and hearing words, similar to how a baby learns. According to Evan Kidd, a psycholinguist at the Australian National University, the study demonstrates that infants may not require specialized cognitive mechanisms to begin word learning. Unlike current chatbots, the new model of language learning is simple and small, deviating from the large language models that rely on massive amounts of data. Wai Keen Vong, a computational cognitive scientist at New York University, highlights that human word learning is not based on the entire internet, but rather on the input provided by parents. Vong and his colleagues intentionally developed a more realistic model of language learning that relies on minimal data to answer the question of whether the model can learn language from such input.
2024-02-12 08:00:00
Source from www.sciencenews.org
