My son asked me a question today that began with school and did not stay there. He wanted to know what I thought about teachers using AI in their work. Should a teacher use it to generate visual aids? What about grading or feedback? At what point does using AI to help with the work become relying on it for too much of the work?
As we talked, he mentioned that he and some of his classmates feel less connected to a teacher who uses AI in a number of ways. I do not know enough about that teacher’s actual use of AI to draw conclusions. Students never see a great deal of the work that teaching requires, and there are legitimate reasons to use tools that make some of it more manageable. What interested me was that he was not really criticizing the quality of what the teacher produced. He was describing a sense of distance from the person teaching him.
Most conversations about AI start from the other side. We ask what the tool can do for us, whether it saves time, or whether it can improve what we produce. Those are reasonable questions, especially when much of our work consists of tasks that simply need to get done. But most of what we produce eventually reaches another person, and that person may experience our use of AI differently than we do.
When the Person Matters Too
There is plenty of work where I care almost entirely about the result. I do not need anyone to spend twenty minutes fixing the spacing in a document by hand when software can do it in seconds. If a tool can sort information or clean up formatting, I am glad to let it. I do not see much value in doing something manually when the tool can handle it just as well.
Some work, though, means more because we know who did it. Feedback from someone whose judgment we respect matters partly because that person actually looked at what we did. A note from a friend means something because the friend took the time to write it. In work like that, the result matters, but so does the attention behind it. For a long time, the work itself gave us some sense of that attention. Detailed comments usually meant someone had read carefully, and a thoughtful message usually meant someone had spent some time deciding what to say. Those were never perfect signals, but they told us something about the person behind the work.
AI makes that harder to read because work that looks thoughtful can now come from very different levels of human involvement. A response may reflect careful thought, or it may have required little more than a prompt and a quick review. That is not necessarily a problem. AI can help someone communicate more clearly, and some tasks never needed much human attention in the first place. But the finished product no longer tells us as much about the process behind it.
What the Result Cannot Tell Us
Imagine receiving a detailed performance review and later learning that your manager had barely looked at your work before asking AI to draft most of the comments. The review might still be accurate, and it might even be better written than what the manager would have produced alone. I suspect many of us would still read it differently once we knew how it had been made, because the information was never the only thing we thought we were receiving. We assumed the review also represented the manager’s attention and judgment.
A message from a friend can work much the same way. Part of its meaning comes from knowing that this person stopped, thought about you, and chose what to say. If much of that process is handed over, the words may still be good, but the exchange changes in a way that is difficult to capture by looking only at the finished message. This is where I have trouble drawing a line. A generated visual may save an hour and make a presentation better. AI may help someone organize their thoughts before writing something difficult. Neither of those uses bothers me much, and I suspect many good uses of AI will look exactly like that.
What concerns me more is how easily the amount of human involvement can shrink without anyone ever making a deliberate decision to step away. Each use can make sense on its own, and over time AI can end up doing more of the reading, drafting, or responding that once required our direct attention. We can remain fully responsible for the work while spending less time with the part another person may have assumed was ours.
Whether that matters depends on the work. If the task is formatting a document, probably not. If someone is asking for our judgment, feedback, or care, the answer may be different because our involvement may be part of what they expected to receive from us in the first place. That is difficult to measure. We can easily calculate time saved or how quickly something was completed. We have no comparable measure for what happens when the person receiving the work feels less connected to the person who produced it, even when the work itself is perfectly good.
I do not know how much is too much, and I am not sure there is one answer. My son was not trying to settle that question either, he was simply telling me what he and his classmates had experienced. They were not arguing that the AI-assisted work was better or worse. They were describing how connected they felt to the person behind it. Maybe that belongs in the conversation too.