This morning a student asked me why, when they create something with the help of AI tools, it doesn’t quite feel as though the result is entirely theirs. I responded like this:
In answer to your question, I think the reason that the project doesn’t feel completely “yours” is that it is not completely yours. To a large extent, when you work with an AI, you are working with the many other people whose expertise has gone into training the AI. You never meet these people, and you don’t know who they are, but the output of the AI reflects their work in addition to your own work. So in a sense you are actually engaged in a collaboration with a very large number of anonymous collaborators. It’s definitely not the same as the experience of building something all on your own.
I’m a little surprised that everyone doesn’t already know this.
By day I work on a very large, very old code base. It is easy to get stuck or puzzled by aspects of it you aren’t intimately familiar with. Pre-covid, you’d get up from your desk and walk over to a colleague’s office for advice. Post covid, mostly working from home, you’d make a Slack post and wait (sometimes for quite a while) for a response.
Now? You can just pose the question to an AI, and let it rummage around the code base and find an answer.
Social aspects of software development, slowly slipping away.