My three-year-old can finish puzzles on his tablet. Hand him a physical one and it’s a different story.
At first I couldn’t understand why. He knew where the pieces went. Then I started wondering what the app was doing for him. Pieces can snap into place. There may be hints, or limits on how they turn. With a real puzzle, he has to turn the piece himself, feel the edges not fit, and try again.
He’s three, so I won’t read too much into it. What got me was how quickly I’d treated a finished puzzle as proof of what he’d learned. I never asked which parts he was doing himself.
And then I started wondering how often I do the same thing.
I think I was better at cursive in third grade than I am now. I learned it, practiced it, and eventually got pretty good at it. Then I stopped needing it. I didn’t decide to forget. Life just offered easier options.
I also remember discovering the synonym feature in Microsoft Word and loving it. I was writing a lot of papers for English, and suddenly I could right-click a word instead of digging through a thesaurus. Sometimes it saved me time. Other times a suggested word helped me figure out what I was actually trying to say. I hadn’t always known what I was looking for until I saw it.
AI is that kind of help, only bigger. I use it to organize messy notes and explore things I wouldn’t have found on my own. Sometimes it shows me what I’m missing. Other times I catch myself accepting something that sounds finished before I’ve worked out what I think.
Searching for the right word can reveal that I haven’t decided what I mean yet. A puzzle piece that won’t fit gives my son a reason to look again. I don’t think every task needs to be harder than it has to be. But I do wonder what we miss when something does the figuring out for us.
Work is no different. When two reports disagree, the first job is to understand what each one is counting. Does “customer” mean someone who signed a contract, someone who was billed, or someone using the product? Both reports can be accurate and still answer different questions. Force them to agree too early and you lose the distinction that mattered. It’s the finished puzzle again: it looks done before the thinking is.
But I’m not dismissing these tools. When my son was sick, a doctor first thought he was having an allergic reaction. I was still worried. I described his symptoms to ChatGPT, and it raised the possibility of staphylococcal scalded skin syndrome (SSSS), a serious bacterial skin condition. It urged me to seek emergency care, which is why I took him to the ER so quickly. The hospital later confirmed the diagnosis.
The doctors and nurses diagnosed him, treated him, and cared for him through several painful days. He’s completely back to his normal self now.
I’ve thought about that experience a lot. I wasn’t looking for an opportunity to figure everything out myself. I needed help recognizing that something might be seriously wrong, and I needed it quickly. The tool gave me information I didn’t have, and I acted on it. I’m grateful it was there.
Maybe that’s part of what I’ve been trying to understand. Sometimes working through something is how we learn. Other times, getting the help we need matters much more than doing it ourselves. The difficult part is knowing the difference.
So I’m not giving up AI, and I doubt I’ll take up cursive again. I just want to get better at noticing what a tool has done for me. Did it help me work out what I think, or just hand me something I was willing to accept?
A finished puzzle doesn’t show me how much my son figured out for himself. And a finished answer doesn’t necessarily mean I understand it.