The views in this post are my own and do not necessarily reflect the views of any organisation that I am associated with.
I feel very uncertain about where we are with AI in practical use. It is unavoidable (literally, every time I open Illustrator, I get prompted to use generative AI) and certainly convenient. I’m also too aware, though, of the flip side of AI to feel very comfortable using it. Indeed, an AI holiday planning mishap figuring out how to see Mount Fuji on a recent trip to Japan led to me and my in-laws taking a mix of local trains, a Shinkansen, and a steep gradient scenic railway, just to end up on the wrong side of a mountain with no view of Mt Fuji at all (remedied by an old fashioned web search and a surprisingly good rural bus service).
But I’ve come across a few places where generative AI has been useful for my work recently, and I’ve started to ask myself a few questions to help decide if AI is the right choice. I’m sure somebody else has given this far more consideration, but it can’t hurt to share these here.
6 Questions to Ask Before Using AI
- Will my use of generative AI displace work that would have been contracted out?
- Will it infringe on somebody’s copyright (or get dangerously close to doing so)?
- Can I trust the output from this task?
- Can I reliably verify the output?
- Where will the output be used?
- Is there a risk of damaging reputations by using AI here?
Basically, will it hurt anyone, can I trust it, and will it make me look stupid? Definitely not complete and I’d love to hear opinions via Mastodon.
So by this measure, I’d feel completely comfortable using generative AI to make images for use in an internal workshop where they can prompt discussion, but not for publication on a company website. I’d feel comfortable generating captions for a video, because the work would otherwise be done manually by me or not at all. I wouldn’t trust generative AI to write a blog post about a complex topic, or to research an area I’m unfamiliar with.