AI agents help me draft some of what I post on GitHub. When one of them actually posts on my behalf, I want you to know. So I’ve started attaching a small signature to those posts:
Generated via Copilot (<model name>) on behalf of @tclem
This is not intended to be an ad. I don’t have an affiliation with any of the model providers and I don’t receive any special compensation for or even measure interaction with this signature — though I do work for GitHub and you should try the GitHub Copilot CLI and GitHub Copilot App!
You might see alternative versions, when, for example, I write something by hand but use a model for light copy editing.
My intention is to provide attribution and breadcrumbs of provenance while retaining a responsible human (me) behind the content. I want whoever reads that content to know how I created it. Just like reviewing code hand-written by a colleague is very different from reviewing code written by said colleague’s AI, review comments, issues, and PR descriptions can be handled differently when composed by a human vs an agent.
I want to be clear and transparent.
NOTE: I do also want to acknowledge that today’s LLMs were trained on public data (and some questionable obtained data) and there are many ethical and legal questions to be sorted out. My opinion is that there must be attribution and credit and ultimately financial compensation of some sort given back to original creators. This is a much larger topic, so consider this a placeholder.
