Skip to main content

Lazy Brain Syndrome and Other Reflections on Using AI

I have a confession. I may be a productive writer, but I’m not an efficient one. I usually start with some jotted notes or a voice memo and rarely have an outline. I clear a path with my own words, often stopping mid-paragraph to jump to another thought. Think of it as a tree. The roots are whatever grabbed my attention. The trunk is my values, the lens I run ideas through. The branches are the paragraphs. Sometimes they arrange themselves nicely, and sometimes one turns into an entirely new branch.

Along the way, I learn. Michael Wald, my co-chair at the Youth Transition Funders Group ages and ages ago, used to say “writing is thinking.”

So I noticed the brain rot the minute I started asking AI to help with my writing. I trained it on my style and was firm that it follow it. I always asked for options instead of “just do it for me.” But gosh darn, it was so much easier to babble my half-baked ideas and instantly get clean text. Pretty soon I was asking for help before I’d written a single word. Bad practice, and a case of what I’ve come to call lazy brain syndrome. As Sydney Schaef and I emphasize in our book Wired for Brilliance (coming early 2027), effortful thinking matters for learning. You gotta work it to build understanding and consolidation (the brain’s way of making learning stick) so you can use it.

Then I had an entirely different experience with Claude yesterday. Yes, it cleaned up some of my text, but what stuck with me was how it pushed my thinking.

I’d picked a topic that had my mind jumping. Seth Godin had named three big capacities (learning, resilience, and commitment) as things managers are responsible for, rather than simply expecting them from employees (and blaming them for failure).

Claude kept taking paths I hadn’t anticipated. It saw things in my writing that I hadn’t seen. Sometimes I loved where it went, even when it was unexpected, and grabbed the text to head down another branch. Sometimes it took my thinking farther than I’d gone. And sometimes it helped me realize I wasn’t going to get where I wanted. It also kept flagging which ideas were its own reasoning rather than established findings, which was a good nudge to go check.

After three versions back and forth with Claude, my big question finally emerged. Each of Godin’s capacities are BIG, and it struck me that integrating them into one organization wouldn’t be easy. Frameworks, vocabulary, and practices are likely to vary. So how much could they actually be integrated? It was an exercise in making sense of multiple change strategies, importance for schools that are transitioning to have the learning sciences at the core of their operations. (Note: I haven’t analyzed whether Godin’s big three are the right ones for schools. And in the public sector, efficiency will always matter when you’re spending public dollars. You’ve got a budget and you have to fit inside it, even if your idea might produce extraordinary achievement gains.)

After writing two articles sparked by Godin’s point, I asked Claude:

I had to do a lot of learning along the way and with your help I kept going in lots of different directions. Can you analyze the “learning path” I followed to get where I landed, and help me think about which questions I’m asking and which ones I’m not? If I understood how my brain sorts things out, I might get better at asking questions that get me to the core of my thinking.

My learning path

Of course it was enthusiastic (isn’t AI always?) and had lots of ideas. First it outlined my learning path. It said I:

  • Have triggers (things I’m reacting to) and walls (the points where I shift directions or turn to AI for help)
  • Test new ideas against my own framework and thinking
  • Get overloaded, then step back and use metacognition (thinking about my thinking) to sort, name the questions, and check my assumptions
  • Triangulate with other perspectives, including my own experiences, other authors, and data
  • Prune and land, by spotting my side paths, linking to real-world experience, and figuring out what the post is really for

The questions I ask a lot, and the ones I don’t

I do a lot of sorting and comparing and contrasting. What I totally avoided is a hard look at where tensions flare between frameworks, or whether any of the frameworks really hold up. Its list of unasked questions was incredibly interesting (although it took me well beyond the blog I was writing).

The most difficult one is being honest about when the AI brings the insight and when I let it slip into my own sense of learning. It’s still my learning, but it wasn’t created by my own default mode network (the brain’s wandering mode) or my working memory. Which, now that I write it, is the lazy brain question wearing a different hat. And with true humbleness, Claude is likely to have different insights than the ones I might have. It also raises a bunch of questions that I’m sure the Silicon Valley AI folks have someone thinking about.

Expanding My Question Asking

Finally, it gave me prompts to carry to my next wall. A few of my favorites:

  • What question is this source answering, and is it mine?
  • What would make me wrong?
  • What’s my evidence, and what’s just a good story?
  • Who’s missing from this picture?

It’s fascinating how helpful a more deliberate routine of question asking could be. (Although the process of discovery is just so fun. Might it take the fun out of learning for me?)

Practices

Of course Claude didn’t stop there. It suggested:

  • Keep a running “learning log” document. For each session, note the trigger (what made you curious), the wall you hit, the questions you asked, the turns that actually moved you, and what you parked for later. Paste the relevant parts at the start of a new conversation and I can pick up where you left off.
  • Add a “which questions did I ask?” line to each entry. After a session, ask me to sort your questions into types (compare, sort, test the evidence, find the tensions, check my assumptions). Over time you’d see your patterns, and whether the gaps I flagged today shrink.
  • Mark who said what. The most useful habit from this process was noticing which ideas were yours, which were mine, and which were borrowed. That’s worth keeping a column for.

Now I can do this myself, or I could switch on memory so Claude remembers our work. I haven’t made up my mind yet. It seems like a lot of work, but it would force me to slow down, which is probably the point. My question asking is rampant, ideas twisting and turning around in my head. My guess is that’s the default mode network wandering and working memory both getting a workout, trying to hold all the threads at once.

Here’s where I’ve landed, for now. The difference between lazy brain and a thinking partner may come down to who is asking the questions. My big takeaway is that I can ask Claude to help me become a better learner and thinker. And if I asked it to show me how it improved my blog articles, it could probably help me become a better writer too.

Over to you: Where are you using AI as a thinking partner, and where is it quietly doing the thinking for you?

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    Resources and insights about how schools and districts are using learning science to create lasting improvements.