The Weirdest Things I've Used AI For

Look, I write about AI tools pretty regularly so you'd think I'm mostly using them for Serious Productive Work Things. And I do use them for that. But I've also used them for some genuinely weird stuff, and I feel like that side of it doesn't get talked about enough. So here's a real list. No judgment on yourself if you see something and think "oh I've done that too."

Settling a soup argument

My friend Priya and I got into this genuinely heated argument about whether French onion soup counts as a meal or a starter. This was a whole thing. We were both very confident. We both had reasoning. I opened up an AI chatbot at the dinner table and asked it to weigh in with actual culinary context.

It gave this really thoughtful answer about how French onion soup occupies this weird middle ground in French culinary tradition, how the caloric density and the cheese and bread element push it toward meal territory in modern contexts even if historically it was positioned as a first course, and how restaurant convention differs from home cooking convention.

Neither of us fully won. But we both felt heard, which I think is the best outcome. And I learned something genuinely interesting about soup. Ten out of ten, would use AI to resolve trivial dinner arguments again.

Practicing a hard conversation

This one I'm a little embarrassed about but I'm including it because honestly it helped. I needed to talk to my landlord about a repair that had been ignored for two months. I am not a confrontational person. I knew what I wanted to say but I also knew I'd get flustered and either undersell how serious the problem was or come off as aggressive when I didn't mean to be.

So I told the AI the situation, the repair, the timeline, what I'd already tried, the outcome I was hoping for, and asked it to roleplay as my landlord while I practiced what I was going to say. I went through it three or four times. It gave me feedback on where I was being too vague, where I was using language that might put someone on the defensive, what I could add to make my position clearer.

The actual conversation went well. I got the repair scheduled within a week. I'm not saying I couldn't have done it without the practice session, but I went in feeling way more prepared than I usually do for that kind of thing, and I think it showed.

Figuring out why my plant was dying

I have a monstera that I've had for four years and I love it in a way that might be disproportionate to it being a plant. Last fall it started looking bad. Yellow leaves, weird spots, generally sad vibes. I'd googled extensively and gotten twelve different possible diagnoses and no clarity.

I described everything to an AI in detail, the specific pattern of the yellowing, where on the leaves the spots were, how long I'd had it, what my watering schedule was, what my apartment light situation was, whether I'd repotted recently, what my humidity levels were like in winter. And it walked me through a process of elimination that was actually really systematic and useful.

Turned out to be a combination of two things: inconsistent watering (I'd been going too long between waterings and then overcompensating) and not enough humidity in winter when my heat was running constantly. Specific fixes, not vague ones. The plant is fine now. Great, actually. New growth. I'm relieved.

Writing a speech for someone else's wedding

My friend asked me to help her write her maid of honor speech about six weeks before the wedding. She knew what she wanted to say but she is not a writer and she was stressed about it. We sat together and I interviewed her, asked her about her favorite memories with the bride, what she wanted people to feel, what she definitely didn't want to include. And then I fed all of that into an AI and asked it to draft something in her voice based on what she'd told me.

Here's my honest take on how this went: the first draft was fine. Not great. It was a little generic in places. But it gave us a real starting point, and from there my friend could point to specific lines and say "that doesn't sound like me" or "I want to say more about this part" and we'd revise. By the end it was genuinely her speech, it had her sense of humor, her specific memories, her relationship with this person.

She delivered it at the wedding and people cried in the good way. I'm counting that as a win.

Naming things

I use AI constantly for naming things and I feel like this is underrated. I've used it to name a newsletter (didn't end up using any of the suggestions but they got me to the right answer through a process of reacting to what I didn't like), name a playlist, come up with usernames for accounts, and name an orchid I was gifted (her name is Veronica and she is thriving).

The orchid naming was the best use of AI I've had in recent memory. I asked for names that felt sophisticated but warm, not too on-the-nose about being a plant name, and had good nickname potential. Got a list of fifty, picked Veronica immediately. Perfect. Fast. Effective.

Talking myself through a decision I'd already made

Okay so this one is a little meta. I had already decided to turn down a freelance project that was paying well but that I had a bad feeling about. I knew I was going to say no. But I kept second-guessing myself because the money was good and I didn't have a concrete reason beyond "I just don't feel right about it."

I talked to an AI about it. Laid out the whole situation. And what was useful was not that it told me what to do, it didn't, but that the act of explaining the situation clearly enough for an AI to understand it made me see my own reasoning more clearly. By the end of that conversation I'd articulated exactly why I had the bad feeling, and it was a real and valid reason, not just anxiety.

I turned down the project. Good call, as it turned out.

I could be wrong but I think this is actually a genuinely useful function, not AI as oracle, but AI as a thing you talk to when you need to externalize your thoughts and don't want to make someone else sit through it. It doesn't judge you. It doesn't get tired. It just asks clarifying questions and reflects things back.

Weird? Maybe. But real talk, sometimes that's exactly what you need.