A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates a novel method for training AI models that significantly reduces computational costs and environmental impact. The technique, called ‘Liquid Neural Networks,’ focuses on creating smaller, more efficient models that are trained on high-quality, curated datasets rather than massive, indiscriminate data scrapes. Researchers found that these compact models can match or even exceed the performance of much larger counterparts on specific tasks like image classification and robotic navigation, while using a fraction of the energy. This approach challenges the prevailing ‘bigger is better’ paradigm in AI development and highlights a path toward more sustainable and specialized artificial intelligence. For the full details, read the complete article at https://technologyreview.com/2024/05/15/mit-liquid-neural-networks.