# Technical writing in the AI Era


Technical writing is very important in today’s AI era. By Technical writing, I mean technical documentation like RFCs, RCAs, etc.

Developers are now writing code with AI, but they have also started writing documentation with AI. 

AI-generated code is fine, as machines read it and can make sense of it. But documents, on the other hand, are read by humans.

I can't read a document created by an AI. It is very difficult for me to read it. It is too big for no reason. AI adds jargon and uses different words in different paragraphs for the same thing. 

Writing documentation is about communicating with peers and other people. 

AI is good for research and finding content for the document, but it's a human job to make it readable. AI writes in a zig-zag flow. It doesn't know about the reader. It doesn't know how the reader will read this document. There is a reader flow.

When humans write, they picture an ideal reader. It can be folks they have worked with, know in real life, and understand their technical capability and vocabulary. 

There were some auto-generated documents before AI, and now I feel they are redundant, like API documentation; most of the time they were just a table of JSON field names and field types. AI can read the source code directly if it is open source, or just read the OpenAPI specs.

I think humans should still write technical documents where readers are human, e.g., RFCs, RCAs, etc. It's more important now than ever. We are shipping more code than ever. We should also make communication with peers more efficient with high-quality documents.