In the realm of artificial intelligence, the ability to generate code has been a subject of intense research and development. Traditionally, AI models have been trained to generate code in a static manner, translating a given instruction from a person into lines of code. I’m sure many of us have tried this on chatGPT with some mixed results… The reason why this approach often fails is because it doesn’t capture the interactive nature of how humans write code. Humans write code in an iterative process, constantly testing and refining their code based on feedback from the execution environment. This is where InterCode, a new framework for interactive coding, comes into play.