Think Together · Start with a Short Story
The robot mistakes a rabbit crouching down for a cat. Instead of only asking “Why are you so silly?” the girl looks for clues it may be missing.
Lay Out the Process
- Data problemToo few examples or bias
- Input problemBlurry or ambiguous input
- Model problemLimited abilities
- CheckFind and correct the error
AI mistakes may come from data, input, model abilities, or a mismatch with the situation where it is used.
Recognizing an unusual pose or handling an unclear request may be outside the model's training coverage. Language tasks can also produce faulty reasoning or made-up details.
When you find an error, saving a specific example and explaining the expected result is more useful than a vague complaint.
Now You Try
If recognition is correct after you replace a blurry picture with a clear one, does that prove the model will never have a problem?
Ready? Open this to see the explanation
No. It only shows that input quality may matter in this case. More tests are needed to understand the model's limits.
One Idea to Take Away
AI mistakes may come from data, input, model abilities, or a mismatch with the situation where it is used.
Try covering the diagram above and explaining each step to someone in your family. If one step is hard to explain, return to that card and look again.