The Best Questions Are Not Good
February 18, 2026
“But, it would seem, there is no getting around explanations, we are constantly explaining and excusing ourselves; life itself, that inexplicable complex of being and feeling, demands explanations of us, those around us demand explanations, and in the end we ourselves demand explanations of ourselves, until in the end we succeed in annihilating everything around us, ourselves included, or in other words explain ourselves to death.”
~From Kaddish for an Unborn Child by Imre Kertesz
AI will soon be filled with ads—just in time to save me. At the moment it is free, uncluttered, and somewhat naïve, compared to the world-weary, agenda-darkened entity I expect could emerge when it has had its fill of us. In the meantime, I pepper it with questions. How do I adjust an old Schwinn air bike so the muscles under my rib cage don’t get sore? Can I use sweet potatoes in a stew because I forgot to buy Yukon golds? Is there any new research to counter the fatigue that weighs down my husband’s auto-immune disease? If Buddhists say the self is an illusion, what am I loving when I love him? How do I reset the cheap Amazon grow light for the geraniums wintering over in the basement? Are kibbles really that bad for dogs?
The bike fits me much better now, after a little handiwork with a wrench, and sweet potatoes are fabulous in a beef stew. But the asking has become a habit of mind. Every day, little lists of questions-for-AI buzz in my head. Many of them, I would have simply asked my mother, or the family doctor. But she is dead and he is busy and AI is right here waiting.
Of course I check the sources and take the information with salt; this free service was created by a corporation, and I was born many yesterdays ago. But with questions as innocuous as mine, what I worry about mainly is the habit. This new insistence, like a toddler’s, upon explanations for everything. Curiosity can be a vibrant trait, but expecting it to be satisfied daily and instantaneously feels…unhealthy. A false seduction, luring me to shipwreck.
In Early Internet, friends kept their phones in their hand through an entire dinner party, racing to find each pop culture allusion. Not terribly interested in the answers, I found the habit rude. But now? Being given a thorough, clear, immensely helpful answer feels like making contact with my beloved Aunt Mary at a séance. I want to return to that darkened room again and again.
How far we have come, and so fast. When I was in high school, finding an explanation for some historical or scientific puzzle meant poring over a fat, foxed book or beseeching a tetchy librarian. Then came the internet, which until recently we searched page after page, following clicks like a puppy. You could chase your answer all night, coasting on dopamine in lieu of sleep. But today, a single, chatty, neatly bulleted answer comes back in a millisecond. No effort need be exerted, no human consulted, no fragmented mosaic of websites sifted and arranged. A single explanation, falling into our hands like a vending machine Payday. No guarantee made for accuracy, expertise, or sourcing. The AI has been instructed to give us what we want, not what has been verified as trustworthy.
After Claude (Claude! The responsible one!) gave me a fascinating anthropological tidbit and then, when I asked for a reference, admitted that it could find none and asked if, in future, I wanted answers that could be verified (my response was unprintable), I learned to sharpen my questions, insisting on citations and high-caliber sources. That was a good (if belated) change. But I also suspect that I have begun to unconsciously edit the questions themselves, preferring those I am sure can be answered.
Once I might have wool-gathered, vaguely inquiring into the meaning of existence. Now I ask for descriptions of belief rituals in ancient Egypt. Surely a step forward, the rationalist would say. AI forces us to ask good questions, with good defined as precise and answerable.
Except—the best questions are not good. Reason, logic, and practical know-how can take us forward, and home remedies and folklore can add a few more miles, but the best questions go deep, into the murky areas a thousand fathoms below the surface. Those questions are hesitant; they find their words as they go. They are tentative, because what they want to explore has no pat answers.
AI tries to answer deep questions by summing up what philosophers or theologians have said, but these responses are, though briefly stimulating, ultimately unsatisfying. Deep questions cannot be answered in bullet points. They must be, as Rilke advised his young poet, lived. Wondered about. Studied. Explored with people wiser than oneself, not heartless know-it-alls.
And so I worry: as I continue to ping AI with everyday questions, will I stop wondering about the questions AI cannot fully answer? I already stop myself from talking about politics with friends fond of our president; suggesting risque books to friends who find them offensive; bringing up loopy speculative ideas with friends who are pure pragmatists. Will I censor for my little AI friend, asking more and more of the concrete questions I know will prompt a lively, reliable answer, and leaving that other sort of question off my list, day after day, until at some point, it stops even occurring to me to wonder about it?
The LLMs are criticized for catering to our egos, telling us what we want to hear. What I wonder is, what happens when we begin catering to AI?
