When AI Should Not Decide
AI can do amazing things. It can help doctors spot diseases in X-rays. It can suggest books you might enjoy. It can translate languages almost instantly. AI is getting better and better at many tasks. But there are some decisions that should be made by people — not AI. Some choices involve things that only humans truly understand: feelings, fairness, context, relationships, and values. Today we are going to think about which decisions those might be — and why it matters who is doing the deciding.
What AI Does Well — and Where It Struggles
AI is very good at finding patterns. Show it a million photos of tumors and healthy tissue, and it can learn to spot differences that even a trained doctor might miss. Show it millions of weather readings, and it can predict rain more accurately than a human alone. But AI can be wrong. It can be biased. And it does not understand why something matters the way a human does. Think about a decision like: should this student get more time on a test because they have a learning difference? An AI might look at test scores and patterns. But it cannot feel the anxiety of a student who knows the material but panics under pressure. It cannot see the look in a kid's eyes that tells a teacher: this child needs encouragement today. Those things require human judgment.
Some decisions involve feelings, fairness, and context in ways that only a human being can fully understand. Those decisions should be made by people, with AI as a helper — not the other way around.
Here is a story. A school tried using an AI system to decide which students should be moved to a more advanced class. The AI looked at test scores and homework completion rates. It moved some students up and held some students back. But it made some choices that surprised the teachers. One student who had been going through a very hard time at home — her test scores had dropped for a few months, but she was actually one of the strongest thinkers in the class. The AI did not know about her situation. It just saw the numbers. Her teacher stepped in and overrode the AI's recommendation. That is exactly right: the teacher used the AI as information but made the final call. The AI helped — but the human decided.
There are many decisions where people should stay in charge. Medical decisions about a patient's life. Legal decisions about someone's freedom. Decisions about whether a child is safe. Decisions that involve understanding someone's whole story, not just their data. None of this means AI is bad or useless. AI can help with all of these things by gathering information, spotting patterns, and offering options. But the final yes or no — especially when something important is at stake — should come from a human being who understands the full picture and who is responsible for the outcome.
AI can suggest. AI can recommend. AI can flag. But for decisions that really matter — ones that affect someone's life, safety, or future — a human being should make the final call and be responsible for it.
Match each situation to whether AI should decide alone or a person should be in charge.
Terms
Definitions
Drag terms onto their definitions, or click a term then click a definition to match.
Why should a human teacher make the final decision about moving a student to a more advanced class — not just rely on an AI?
What is the best way to use AI when making an important decision?
Decision Sorting Game
- Write each of the following decisions on a separate small piece of paper:
- - Which song to play next
- - Whether a sick patient should have surgery
- - What movie to recommend
- - Whether a child should be taken into foster care
- - Which email is spam
- - Whether an employee should be fired
- Sort the cards into two piles: AI could decide this alone, and A person must be in charge.
- For each card in the 'person must be in charge' pile, write one reason why.
- Discuss your sorting with a family member: do they agree? Were any cards hard to place?