Six ordinary things AI helped me finish or understand this week

I used AI six different ways this week: to cook dinner, build a private game tracker, understand a giant education study, assemble an audio project, examine how I deliver criticism, and bring the Dear Future Overlords website and Substack back into agreement. Several results are already in use. The audio still needs my ears, and the communication work gave me something to think about rather than a tidy ending.

Frozen chicken became dinner

I gave AI the frozen chicken, vegetables, and condiments I actually had. Its first cooking time was wrong, and it forgot to tell me to cut the vegetables. I corrected both and kept cooking. Dinner was pretty good; next time I would use more sauce.

What I got out of it: Dinner from food already in the house, without handing the entire job to Jason.

Why AI was useful: It kept track of the ingredients and sequence, stayed with me while I cooked, and revised the plan when my household evidence was better than its first answer.

A private game tracker replaced a mental checklist

I turned a large Phantom Weapon checklist into a private website for Jason and me, with separate progress, character images, weapon stages, and draggable priorities. Once I used it, I found problems with completed-item sorting and drag scrolling. AI fixed both and redeployed the tracker.

What I got out of it: A private tool shaped around the way Jason and I actually approach the game instead of a spreadsheet I would eventually ignore.

Why AI was useful: It carried a weirdly specific idea through game research, design, coding, private sign-in, and deployment. Conventional software was never going to fit us this closely.

A giant education study became one usable sentence

The first summary of the OECD's PISA 2025 results was too neat, so AI and I went back to the source. The line I kept is: Using AI isn't what seems to matter most. Knowing how to use it does. I have not tried it in a class yet.

What I got out of it: A more accurate understanding of the study and language I can plausibly use in an introductory class.

Why AI was useful: It read enough of the primary material to catch the overstatement, keep the uncertainty intact, and help me say the result in language a normal person might willingly hear.

A repeatable Audacity project removed about two hours of assembly

AI assembled and mixed a Dear Future Overlords transmission in Audacity, saved the project, and documented how to repeat the work. It should save roughly two hours per episode. I have not approved the final product; it still needs my listening review.

What I got out of it: A fully assembled Audacity project and a repeatable process that should save roughly two hours per episode, while leaving the final listening decision with me.

Why AI was useful: It handled the exact, repetitive work inside Audacity and wrote down the process as it went, so I should not have to rebuild the same two-hour setup by hand for every episode.

I found the question I was not asking before delivering criticism

When a problem needs fixing, I can focus on it so completely that I miss what accepting my criticism may cost the other person. Looking across several private conversations gave me a useful question: What might they think I am saying about them?

What I got out of it: A clearer picture of my own communication pattern and a practical question to carry into the next difficult conversation—not a tidy resolution.

Why AI was useful: It could look across several private conversations and spot the attention shift I could not see while I was inside any one of them.

Claude made the DFO website and Substack describe the same show

The DFO website and Substack were telling visitors two different stories about the same show. Claude helped rebuild the website's paths into the work and made 35 verified changes across Substack, including its description, navigation, subscription language, emails, podcast details, and episode numbering.

What I got out of it: The DFO website and Substack now introduce the same listen-first show, make the same promises, and give a new visitor a clearer way to begin.

Why AI was useful: Claude could keep the whole mess in view across two systems, trace old language into places I would not have thought to check, and revisit its own conclusions when the public result proved them wrong.