AI란 무엇인가?

AI is an umbrella term: machine learning is one approach within it, and generative AI is a type of system that produces new content.

주요 시사점

• AI, machine learning, and generative AI are related but different terms.
• A convincing output is not proof of understanding or correctness.
• Judge a system on the task it must perform and the consequences of its errors.

An AI system takes inputs, processes them using rules or a learned model, and produces an output. A route planner might search possible journeys using explicit rules. A machine-learning model might estimate a delivery time from examples of earlier deliveries. Both can be useful without thinking or understanding in the human sense. The distinction is how the system reaches its output, not whether its interface looks intelligent. Machine learning replaces some hand-written decision rules with patterns learned from data. Generative systems use learned patterns to produce text, images, audio, or other outputs. A chatbot can therefore produce a fluent explanation without checking whether every statement is true. Its ability to generate a response is different from evidence that the response is correct. To evaluate an AI claim, identify the task, the input, the output, and the evidence used to judge success. A good result on familiar examples is not enough: ask what happens with unfamiliar data, ambiguous requests, and costly mistakes. Human review, clear limits, and a way to challenge an output matter as much as the model's headline capability.

1. A rule-based sorter sends every message containing the word 'refund' to a billing queue.
2. A learned classifier is trained on messages that people have already labeled as billing, technical support, or general questions.
3. Test both on fresh messages, including 'I do not want a refund; I need help logging in.' Count incorrect routes and review the costly mistakes.

The rule and the classifier can fail differently. This illustrative comparison shows why the label 'AI' alone cannot tell you which system is more useful.