On a recent college visit with my daughter, the University of Indianapolis sent the prospective students off to sample classes and gathered the parents for a Q&A. We could ask about the things we wanted to understand before sending our children there. Somewhere in that conversation, AI came up.

We heard about a peaceful student demonstration connected with the university's AI Summit earlier in the year. The school had brought business and technology leaders to campus to discuss AI and its use. Some students had come out to protest.

I wasn't at that event, and during the Q&A I didn't know what the students' specific objections were. Organizer Cassidy Rexroad told Bloomberg:

Doing the work for yourself has value. Having the human connection has value.

I use AI every day. I also want my daughter to get an education worth having. Those two things give me a reason to listen to that objection.

Some of my AI use is fairly ordinary. I ask for help planning a garden or working through a recipe for ribs or brisket. I use it to help organize family devotions and prepare apologetics material. At work, it helps me write and review code, investigate problems, and build features that would take more time to develop on my own. I find it useful enough that I would recommend exploring it, with some care about what you trust it to do.

I have been developing software since the mid-1990s. That experience helps me judge a proposed design or recognize code that needs a closer look. It does not make me qualified to accept every answer AI gives me on every other subject. Familiarity with the software can make that distinction easy to forget.

My Family Devotion is one reason I take the benefits seriously. I built it to help families, including mine, read and discuss the Bible together. AI assists with drafting and research. The work also includes checks against the Berean Standard Bible, verification of outside claims, and my own editing and review before publication. Families get a free resource with the passage, discussion questions, an application, and a prayer.

AI helped me make that resource available. A Christian purpose would not excuse a fabricated quotation or a careless explanation of Scripture. If I use AI to prepare an apologetics argument, I still need to read the sources and represent the opposing view fairly. Producing more material would be a poor result if I made it less trustworthy.

At MySafeInfo, I built an AI worker that revisits source pages on a schedule and proposes updates to reference datasets, as well as researching new ones. That takes some of the repeated collection work off my plate. The publishing process requires source evidence for every accepted value. Those checks can still miss mistakes, but they give me a way to reject an unsupported suggestion before someone uses it.

Education adds another complication. Sometimes doing the work yourself is the purpose of the work. A student who submits a correct answer may still have missed the practice that would help them solve the next problem. I spent years learning to recognize mistakes in software. I would be uneasy about recommending a way of learning that let someone skip all that practice and then expected them to supervise the machine.

A study of nearly a thousand math students at one Turkish high school makes the distinction concrete. Students using a general-purpose AI tutor did better on practice problems than students without AI assistance, but worse when both groups took exams without it. A tutor designed with teacher input to guide students rather than hand over answers largely avoided that penalty, though it did not establish better unaided exam performance than the control group. That is one study in one setting, not a verdict on every classroom. It does show why finishing an assignment and learning from it need to be measured separately.

Then there are objections that cannot be answered by showing someone a well-checked piece of work:

• AI has physical costs. The International Energy Agency's 2026 assessment describes growing electricity demand from AI-focused data centers even as individual tasks become more efficient. Power supply, local infrastructure, and who pays for expansion are reasonable things to ask about. Checking an AI-generated answer does nothing to settle those questions.
• Workers may experience the same efficiency very differently. Time saved can mean room for better work, but an employer can also use it to cut staffing or increase workloads. My experience as a developer does not tell a student what opportunities will be available when they graduate. I cannot promise them that everything will work out because technology has changed jobs before.
• Writers, artists, and other creators have a stake in how their work is used. Questions about permission, credit, and compensation deserve an answer. Finding an output useful does not establish that everyone whose work contributed to the system was treated fairly.
• Other people's information is not mine to hand over casually. A client document, a student's record, or a private family conversation needs more consideration than whether pasting it into a chatbot would save time. Responsible use includes deciding what should never go into a particular service.

I do not have a complete answer to all of that. I also do not think the choices are limited to giving up AI altogether or defending everything its providers do. I can use a service, ask harder questions of the company behind it, and decide that some applications are unacceptable. Schools and employers have decisions to make that an individual user cannot make for them.

When a school adopts AI, I want teachers and students involved in deciding where it belongs. When a business uses it for a decision that affects someone, that person should have a way to get an explanation and challenge a mistake. A system that saves its operator time by making everyone else work harder to correct its errors is difficult to defend.

I intend to keep using it. For someone who is curious but hesitant, I would start with a task they already understand well enough to check. Keep private information out, and count the time spent reviewing and correcting the answer when deciding whether it helped. There is no obligation to use it for the next task just because it worked for this one.

As a parent considering colleges, I want my daughter to learn to work with AI and to question it. I want her teachers to know when assistance helps and when she needs to struggle through something herself. I would like to hear how the college makes those decisions, and how its students help shape them.