How an astrophysicist uses Codex to help simulate black holes

According to an OpenAI Blog post, astrophysicist Chi-kwan Chan has been using Codex, an AI code generation model, to build simulations of black holes. Chan, who studies extreme gravitational environments, leverages Codex to write code that models accretion disks, magnetic fields, and other complex phenomena around black holes. This approach allows him to rapidly prototype and iterate on simulation scripts, accelerating the testing of theoretical predictions from general relativity. The article highlights how Codex can understand domain-specific scientific requirements, translating high-level physics concepts into functional code. For developers building AI workflows, this case study demonstrates the potential of code-generation tools to handle specialized, non-trivial tasks—not just boilerplate code. It shows that AI can be a collaborative partner in research computing, reducing the time from idea to simulation.