Last Saturday I found myself at a Goodwill with about twenty minutes to kill. I wandered over to the book section, which was the usual mix of old romance novels, outdated textbooks, and the occasional gem hiding in plain sight. The problem is always the same: there are hundreds of books crammed onto those shelves, and I never know which ones are actually worth the dollar or two they're asking. So I tried something new. I pulled out my phone, snapped photos of a few shelves at a time, and fed the images into an AI assistant. I asked it to tell me which books were well-regarded, which ones were still relevant, and which ones I could safely skip. Within a few seconds I had a shortlist with brief explanations for each recommendation. It worked surprisingly well. The AI flagged a copy of "Thinking in Systems" by Donella Meadows that I would have walked right past. It also told me that a data analytics textbook from 2011 was too outdated to bother with, which saved me from a purchase I would have regretted. I ended up walking out with four books for six dollars, and every one of them has been a good read so far. What I liked most about this wasn't the money saved. It was the speed. I went from "overwhelmed by a wall of spines" to "confident in my picks" in under five minutes. That's the kind of thing that makes AI feel genuinely useful to me: not replacing my judgment, but helping me move through decisions faster when I don't have deep expertise. I've been thinking a lot about how [[signal/AI Tools for Analysts|AI tools for analysts]] work in professional settings, but honestly, the thrift store might be where I felt the value most clearly. It reminded me of some patterns I picked up in my [[notebook/Coursera AI Foundations|Coursera AI notes]] about using models for quick classification tasks. Turns out a pile of used books is a perfectly good classification problem. ← [[Life|Back to /life]]