However, they are currently limited to executing only those tasks they are specifically designed to perform and are unable to adapt when encountering situations outside their programming or training. DARPA's Lifelong Learning Machines (L2M) program, drawing inspiration from biological systems, seeks to develop fundamentally new ML approaches that allow systems to adapt continually to new circumstances without forgetting previous learning.

Above- Today's machine learning and AI systems are limited to executing only tasks they are specifically programmed to perform, without being able to adapt to new situations outside of their training. DARPA's L2M program aims to generate new methodologies that will allow these systems to learn and improve during tasks, apply previous skills and knowledge to new situations, incorporate innate system limits, and enhance safety in automated assignments.