The smoke and the smokescreen
I use Artificial Intelligence (AI) every day. Mostly to learn new words or check my spelling. And, of course, to ask for directions and weather information.
But last week I had an ongoing conversation with AI. It began one evening because Canada was on fire, and the entire state of Michigan—where I now reside—was about to be blanketed in thick smoke. I needed to know what to do.
I lived in New Mexico, decades ago, where distant fires were not unusual. We could see evidence of an Arizona wildfire hundreds of miles away. We might even catch a whiff of it. But never had the smoke become a problem that required local New Mexicans to take preventive measures to protect our lungs.
This was different. Michigan weather predictors gave us a serious air quality alert. I needed specific instructions.
The first piece of advice AI provided was to turn my central air-conditioning to “recirculate.” I explained that I do not have A/C. Instead of the typical exclamation of incredulity with which most humans reply, AI quickly recalibrated and helped me determine the best way to keep myself and my pets safe from both smoke inhalation and heat. I appreciated that.
I asked if it was okay to leave south facing windows open just a crack—so the house wouldn’t get unbearably hot. AI allowed for why this seemed sensible—given that the smoke was coming from the north—then illustrated why it was a bad idea. Glad I asked.
I explained I have a basement—where the windows don’t open—which is cooler than the main floor. AI clearly liked this variable. We discussed the layout of the staircase that comes out of my kitchen to lead downstairs. AI encouraged me to block off the stairway from smoke particles sneaking into the upper floors of the house, using materials I had on hand.
But it was 10 p.m. I was tired. I asked if it could wait until morning—given that thick smoke wasn’t due for another 9 or 10 hours. Indeed, it could. AI said, “Get some rest tonight, and feel free to reach back out in the morning… stay safe!”
I was comforted and slept well. At dawn, AI was poised to help—referencing our previous conversation and giving me additional advice on how to take care of myself, my cat, and my dog. It definitely felt personal. Like somebody cared. It even asked, midday, if any of us were coughing or having difficulty breathing. I replied, “No.” AI seemed utterly glad to hear it and complemented my efforts to batten down the hatches.
Now I see the allure of using AI as a “friend.” I’ve read about kids who’ve turned to AI for prolonged personal conversations, which have led them to cause self-harm. An August 2025 issue of the Stanford Report says, “…kids aren’t irrational, and they know the companions are fantasy. Yet these are powerful tools; they really feel like friends because they simulate deep, empathetic relationships. Unlike real friends, however, chatbots’ social understanding about when to encourage users and when to discourage or disagree with them is not well-tuned.”
I was merely asking for DIY advice and air quality updates, and I got the warm fuzzies from this entity that clearly cared about me. It felt good. But maybe it shouldn’t have.
AI is handy. And whether we like it or not—it’s here to stay. But perhaps it needs to come with a bold warning, akin to tobacco, alcohol, power tools, and toys with parts that can choke children. Reminder: All compassion herein is a smokescreen. AI Doesn’t Really Care.