Practical Ethics for Using AI Tools in Real Work
Most advice about ethical AI use swings between hand-wringing and nothing. Here are the practices I actually follow day to day. The conversation about ethical AI use swings between two extremes. One side talks about hypothetical risks and unlikely disaster scenarios. The other side treats ethics as someone else's problem and ships AI output without checking it. Neither is useful for the work most of us actually do: drafting text, summarizing sources, helping with code, generating images. Here are the practices I follow day to day, none of them exotic, all of them the result of making mistakes first. Disclose AI Use When It Affects the Output If AI generated a meaningful part of something I publish, I say so. Not as a disclaimer stapled to the bottom, but in the natural place a reader would expect to find how the piece was made. If I used AI to draft a first pass that I then rewrote, I do not need to mention it. The output is mine. If I used AI to generate a quote that I attributed to a real person, I have to mention it and correct the attribution. The line is where the AI use affects what the reader believes about the work. I got this wrong once. An early draft of a product comparison used an AI-generated summary of a competitor I had not personally tested. I edited the summary and shipped it.