New image model from OpenAI!

I went back to my hometown McDonald's last week. It made me rethink how I talk about AI. I grew up in this town. I know this McDonald's. And last week I walked in and it was gutted — no cash registers, no one up front. Just a row of digital kiosks. The actual workers were pushed back into the kitchen, tucked behind a corner where you could barely see them. Then they forgot one of my fries. Before, that's a 10-second fix. Walk up to the counter, say something, done. Instead, I had to walk behind the counter, into the kitchen, and try to flag someone down while everyone back there is slammed. I felt like I was doing something wrong just by being there. A problem that used to take zero effort to solve now takes effort, awkwardness, and a little bit of guilt. That's not AI making my life easier. That's AI making my life harder so a company can save on labor. And I think that distinction matters a lot. Here's where I land on all of this — I'd call myself a realistic optimist, or a cautious optimist. I'm genuinely bullish on where this technology goes. I think AI has the potential to free up a massive amount of our time — time we could spend picking up new hobbies, sharpening skills, learning things we never had bandwidth for, and actually showing up for our communities instead of being buried in busywork. I use it every day to think out loud and organize my own head. That part's real, and it's good. But I'm also watching it get implemented in ways I don't agree with, right now, not hypothetically. The McDonald's kiosk is a small example of a bigger pattern: technology getting rolled out to cut cost or friction for the business, at the direct expense of friction for the human on the other side of it. We should be way more intentional about that trade-off than we currently are. The other thing I'm protective of: our ability to think for ourselves. AI is an incredible tool for brain-dumping, organizing, drafting — I lean on it constantly. But there's a real difference between using it to support my thinking and letting it start replacing my thinking. That line is worth guarding, especially as it gets easier to just let the model decide.