A student can ask a chatbot to solve a problem and then write the explanation a professor expects to see. If the assignment also calls for a reflection on what they learned, the chatbot can write that too. So when that submission reaches me, I have to decide how much it tells me about what the student understands.
Dr. John Delano and I spent much of this year working on COACH, our approach to teaching and assessing student judgment when AI is involved. The paper we wrote about it received a best paper award at the Christian Business Faculty Association conference this fall. The paper was written for professors, but three of its lessons have influenced how I make decisions of my own.
1. Using AI well takes practice with feedback
After a few months with a chatbot, most of us learn which prompts get useful responses, and we grow comfortable with it. What we rarely do is go back and check how the decisions it helped us make actually turned out. The vendor you picked on its advice, the plan you presented at work, the reply you sent in a tense moment: did the reasoning hold up? A survey of knowledge workers found that the more confident people were in the AI, the less critical thinking they did with its output (Lee et al., 2025). Comfort lowers our guard.
Judgment improves with practice and feedback, like any other skill. Abrami and his colleagues (2015) found the biggest gains when people worked through realistic problems and talked their reasoning over with someone who could guide them, and Halpern (1998) stressed practicing deliberately while paying attention to our own thinking. My suggestion is simple. For a decision that matters, keep a short note of what you accepted from the tool, what you rejected, and why. Once you know how things turned out, go back and read it. A good outcome can owe a lot to luck, so look at how you reached the decision and not just how it ended. Without a record, most of us remember our reasoning more generously than it deserves.
2. Ethical concerns should influence what we decide
Ask a chatbot about a plan to cut costs, and the answer may note that an employee loses hours and still advise you to go ahead. Reading that acknowledgment can reassure me that the advice has properly considered the other person’s situation. It hasn’t, necessarily. A list treats every item as the same kind of thing. The savings and the person sit side by side, and nothing in the format says which should carry more weight. I may accept the recommendation before deciding whether that person’s situation should have changed it.
Our paper examines how an ethical concern can appear in a recommendation while having little influence over the proposed action (Leo & Delano, 2026).
Micah 6:8 asks us to do justice and love mercy as we walk humbly with God, and Jesus made it personal: love your neighbor as yourself (Matthew 22:39). A neighbor is not a line item. That gives us a standard for evaluating advice, especially when the option that suits us best would come at someone else’s expense. So before you follow a recommendation like that, I suggest three questions. Who does this affect, and how much? Would another approach treat them better? And are you willing to take that approach when it costs you time or money? A chatbot can name what justice requires. It cannot love the person on the other end of the decision.
3. Explaining it out loud is the one thing AI can’t do for you
Anything written can now be generated, including the explanation of why a decision makes sense. A conversation can’t. It happens in real time, in your own voice, in front of someone who can ask the question you didn’t see coming. A good question brings reasoning to the surface where it can be examined, and it shows whether there was any reasoning there to begin with. That is why our paper relies on brief spoken explanations, and not only written work, to assess judgment.
Romans 14:12 reminds us that each of us will give an account of ourselves to God, and we are still responsible for the decisions we make with AI’s help. So before you act on advice that could seriously affect another person, I suggest you explain it to someone willing to push on your reasoning. Their questions may turn up an assumption you accepted too quickly or a consequence you underestimated, and it is far better to hear that while there is still time to reconsider.
References
Abrami, P. C., Bernard, R. M., Borokhovski, E., Waddington, D. I., Wade, C. A., & Persson, T. (2015). Strategies for teaching students to think critically: A meta-analysis. Review of Educational Research, 85(2), 275–314. https://doi.org/10.3102/0034654314551063
Halpern, D. F. (1998). Teaching critical thinking for transfer across domains: Disposition, skills, structure training, and metacognitive monitoring. American Psychologist, 53(4), 449–455. https://doi.org/10.1037/0003-066X.53.4.449
Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025). The impact of generative AI on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (Article 1121, pp. 1–22). Association for Computing Machinery. https://doi.org/10.1145/3706598.3713778
Leo, A., & Delano, J. D. (2026, October 22–24). COACHing students to think beyond the algorithm: A framework for AI-mediated judgment [Conference presentation]. Christian Business Faculty Association Annual Conference, West Palm Beach, FL, United States.