I used AI to make two novels that only I would read. Then I read them. Parts were bad. The characters sounded too much like one another, and I gave the result about four out of ten.
I wanted to keep going anyway.
That sounds like a strange opener for a list of things worth keeping. But it explains the difference between this episode and the regrets. Sometimes I use AI and skip an experience I wanted. Sometimes it gets me into an experience I wouldn’t otherwise have tried.
1. Making the frivolous thing
The novels were an indulgence. Then they became a new interest.
I’m a finance and data person, so I started comparing their structure with books I enjoy: sentence length, paragraph length, the share of dialogue. I could see patterns worth experimenting with. I could also see the limit of the measurements. A dialogue percentage didn’t tell me whether the characters had distinct voices.
I went looking for a book and found something I wanted to learn about. That’s why this stayed on the keeper list, even though the first output disappointed me.
2. The other person in the room
Working alone, I often want someone to challenge an idea or rehearse a conversation with me. AI can play that role when nobody is available at the next desk.
I have to ask for disagreement. I might ask it to make the strongest case against my plan, or play a contractor so I can practice explaining what I need. The objections are things to examine; the simulation doesn’t tell me what the actual person thinks.
3. Walking in prepared
Before a complicated appointment, I can work through the paperwork and prepare questions. The useful result is a better conversation with the person I’m meeting.
I still need them to confirm the interpretation. The AI’s job is to help me notice what I don’t understand and arrive with a short list of questions that matter.
4. Building small things for myself
I’d tried learning Python and Swift. The gap between starting and getting something useful was long enough that I lost momentum.
AI shortened that gap for the kinds of personal tools I wanted to make. It didn’t make me a software engineer. It gave me a way to build a first version, try it, and learn from what happened. My notes system is one example of the kind of personal need I can now work on directly.
5. Being interviewed about a change
A list of goals is easy to request. Understanding why I keep circling a change takes more work.
I find it useful to ask for an interview, one question at a time: what do I want, what have I tried, and what got in the way? I treat its interpretations as possibilities to correct. I get to decide what fits.
Find your keeper
The companion exercise gives you five small ways in. Choose one and check whether it earns a repeat. You don’t need five new workflows, or two novels, by the end of the week.