That’s the ultimate goal of anything informed by Artificial Intelligence (AI), whether writing a school essay, offering customer support, scanning a medical image to create a diagnosis, or simulating the behavior of a system. But AI is massively misunderstood right now. We’re caught between the science fiction of “machines that will take over the world and destroy humanity” and the reality of where we are today. It’s time to step back and see how AI –and, more realistically, Machine Learning (ML) – is helpful in simulation today.

Let’s start with ChatGPT, which created a media storm in late 2022. ChatGPT was developed by OpenAI, a company that wants to build and productize AI systems that can be trained to interact with humans. OpenAI isn’t the only entity working on this; it has many academic and commercial partners and competitors. They’re all trying to solve the problem of building technologies that can receive inputs (all of Shakespeare’s sonnets, let’s say) and use that to create useful outputs (so not just reciting Shakespeare, but writing original verse). According to OpenAI, “ChatGPT will someday be able to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.”