Today’s AI models are typically designed to accomplish one task, requiring us to train a new model from scratch whenever we need it to do something else. Pathways is an innovative approach that allows us to train a single model capable of successfully completing thousands or even millions of tasks. Rather than starting with random numbers, Pathways enables the machine learning models to build on their existing knowledge and skills when tackling new tasks. This means that what was learned by training for one task – for example recognizing aerial images and predicting elevation – can be applied and combined in order create more complex processes such as predicting how floodwaters will flow through terrain. The end result is faster learning times and improved accuracy. In addition, Pathways also allows for multiple senses within a single model, meaning it can take into account both audio cues and visual information when making decisions – similar to how our brains naturally process input from different sources simultaneously.