Recently I investigated a few AI-powered tools for writing developer documentation. My investigation was prompted by two things:
I had a front-row seat in GPT’s popularity rise; I documented a GPT-3 powered tool from my client Sensible back in December 2022, just before GPT’s explosive entry into the public imagination, and I was immediately stunned by its potential compared to other AI tools I’ve documented in the past. So I was pretty excited to see what large language models (LLMs) could do for generating documentation.
Existential dread curiosity – will AI replace documentarians? My initial answer is no; we still need humans to write high-quality developer docs, but tools like GPT will absolutely change how we write. And I think that’s a good thing! I’m encouraged by early research on AI’s transformative role, such as those that found that “separate studies of both writers and programmers find 50% increases in productivity with AI, and higher performance and satisfaction”.
I didn’t realize that the “explain” feature is part of Copilot Labs. I signed up for Copilot, then installed both Copilot and Copilot Labs in VS Code from the Visual Studio marketplace, and made sure that I’d authorized Github in VS Code by previously installing the GitHub Pull Requests and Issues extension.