The computation revolution started when we built general purpose computers to follow instructions. A computer follows an algorithm step by step. It doesn’t need to understand the algorithm.

Here’s an example to illustrate the difference between understanding something and following an algorithm. There’s a world renowned chef, Andre, who makes a special pasta dish. Andre lost both his arms (sorry Andre). He hires you to make this dish. You don’t know anything about how to make this dish. You don’t know these ingredients, how they taste, which ones blend together. But, Andre is great at communication. He gives you step by step procedure to make the dish and you follow along with him. Voila! You made the special pasta only Andre knows how to make. Andre understands how to make this special pasta. You followed an algorithm. But what is it that Andre has that you don’t? Context. He knows the flavour of the ingredients, he knows what they taste like, he knows which ingredients complement each other and which don’t. He has context around the algorithm. This context is what I proxy for as understanding.

The AI revolution is about giving computers more context around the computation they perform.

“Write me a function to add five numbers.” - you give the prompt to a language model that runs on a computer. The computer doesn’t have an algorithm to give you the output you want. It only knows how to execute an algorithm. The language model contains the context to make head and tails of the prompt. We will provide this context to computers via AI such as natural language, our physical reality, our visual imagery, anything and everything we have access to. With time they will have as much context as we do. This is what I call the Context Revolution.