Large language models have hit the scene and they represent a large disruption to how work is traditionally done. Programming is no exception.
I’ve been using LLMs like chatGPT and copilot for almost a year now. I saw potential of using the tool for my own creator activities. I have some thoughts on how AI may change the profession of programming.
As a quick aside I summarize my LLM usage in two broad categories: strategic versus tactical. Strategic questions represent research and design; chatGPT is excellent at serving this role. Tactical questions represent how to create or manipulate software to do what I want to do; both chatGPT and copilot fulfill this. I believe I am the entity that wants and the LLM tooling dutifully helps me get what I want.
Writing software to interact with AI APIs directly opens up new possibilities. What happens when I could create a solution (or system?) where bots talk to each other. Specifically I want to divide up the labor of creating video content to multiple AIs, each serving a role in the video delivery. Perhaps I have a bot that thinks up a list of content ideas and remembers which ones are chosen. Another bot thinks up scenarios that fits the content idea. Yet another bot writes a script, another bot edits the script to tighten it up, etc.
At some point I, the human, will intervene in the creation of the work of art to leave my own mark but ideally things are automated.
Does this make me some sort of middle manager or executive of a bunch of robots? I suspect so.