WHAT ARTIFICIAL INTELLIGENCE (AI) IS AND IS NOT

CHAPTER SUMMARY Artificial Intelligence is the practice of building systems that handle tasks which seem to need intelligence, often by learning patterns from examples rather than following hand-written rules. Narrow AI does one kind of task or a family of related tasks, and every deployed system today is narrow, while general AI remains a research goal. The history runs from rule-based systems, through machine learning, to deep learning and the transformer-based language models behind modern assistants. An agent is narrow AI arranged so that it can pursue a goal over many steps using tools, and its reliability depends on the software around it. KEY TERMS REVIEW AI: systems that perform tasks that seem to require intelligence. Narrow AI: AI built for one task or a closely related group of tasks. AGI: the hypothetical ability to learn and do almost any intellectual task a human can. Rule-based system: software that follows rules written by people. Machine learning: improving at a task by studying examples. Deep learning: machine learning using neural networks with many layers. Neural network: a structure of many linked numerical units whose link strengths are adjusted during learning. Agent: software that pursues a goal through repeated actions and observations. Deterministic code: code that gives the same output for the same input.