A new study from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) demonstrates a novel approach to AI training that significantly reduces computational costs. The method, called ‘Learn to Learn to Learn’ (L3), enables AI models to build upon previously acquired knowledge much more efficiently than current transfer learning techniques. Researchers report that models trained with L3 required up to 50% less data and computational resources to master new, related tasks compared to standard methods. This advancement could lower the financial and environmental barriers to developing sophisticated AI, making powerful models more accessible to researchers and organizations with limited resources. The full details of the study are available in the published paper: https://www.technologyreview.com/2024/05/15/1092671/mit-ai-training-method-cuts-costs/