Generally speaking, the more parameters that a Large Language Model (LLM) has, the faster it learns. There are exceptions. For example, Google’s original PaLM LLM had fewer parameters than PaLM 2, however PaLM 2 learns more efficiently than its predecessor.
LLMs are of great importance given that they underpin Gen AI applications and services. The next generation of Gen AI chatbots will be powered by ever more powerful LLMs. Think of Google Assistant on steroids – a version that can order your meds and remind you take them, a version of Assistant that can prescribe the optimal diet (including recipes), based on your real-time health markers, one that can order takeout, book flights and hotels, execute financial transactions, execute an Amazon shopping spree and more. It won’t mow your lawn for you, but it can book a landscaping service and write a Google Review.