kate mors
5 min read
artistic doodle
artistic doodle

The interactive encyclopedia, human accountability, and cross-checking every answer

Deniz

Finance

ChatGPTClaudeVoice Mode

You said that you fell in love with AI the first time you used it. Do you remember what task or prompt it was?

It wasn't any task in particular, just the idea of having something that holds all the knowledge I could ever need. It gave me a good feeling, knowing that I could ask it anything. It's like having every encyclopedia I could ever want and getting an answer right away.

Can you walk me through a specific workflow where you'd actively miss AI if it disappeared tomorrow?

I don't think I would miss it in the sense that I wouldn't be able to function without it, though it is comforting to have that knowledge at my fingertips. In my daily work, I use it to double-check my own thinking and make sure there is nothing I overlooked. It often points out an angle I hadn't considered on my own.

“If I just never get back to it [AI], nothing is going to happen. It won't break the relationship, and it will respond to me the same way as before.”

If you could redesign one AI tool you use daily and change one thing about how it works, what would that be?

I use voice mode very often, even for pep talks, and I've been thinking about this question a lot. I don't think what I wish for is possible because it comes down to human nature. When you form a human connection, you create trust and the accountability that comes with it. In a friendship or relationship, if you break trust, you have to mend it and show up. How AI could ever handle that piece is something I think about a lot.

Everyone talks about AI "saving time". Realistically, when AI saves you an hour, what happens to that hour?

I never consciously thought about what I do with the time I save. I use that time to do things AI cannot do. Instead of spending twenty minutes wording an email, I let AI draft the email and go empty the dishwasher, which AI definitely can't do right now.

When you are writing a prompt, what is your mental model of who you are talking to?

Image-wise, I don't have a man or a woman in mind. I think that's partly because Turkish doesn't have grammatical gender, so I don't picture a person. I think of it as an interactive encyclopedia that I'm bouncing ideas off. It doesn't have a face in my head.

If you had to teach a colleague how to get the same value out of AI that you do, what is the first habit you would teach them?

I would tell them to build on the conversation step by step instead of just asking a one-off question. Treat it as something you can think with, rather than an advanced Google search.

“Computer scientists are comfortable iterating again and again, whereas I expect a usable answer from one prompt.”

What is the "hard line" task that you would never let an AI do on its own?

In finance, precision is the hard line. I have AI do the first pass on simple tasks, and then I check every single answer. Most of the time it's right, but not always, so I verify the numbers before I use or send anything.

Is there something you would like to learn more about AI?

The coding part doesn't interest me at all. What interests me is training agents, though I don't have the patience for endless trial and error. Computer scientists are comfortable iterating again and again, whereas I expect a usable answer from one prompt. Learning a practical way to set up agents without all that repetition is what I want next.