Introduction
Artificial intelligence, usually shortened to AI, is a broad field of computing concerned with building systems that can perform tasks involving aspects of intelligence. These tasks can include recognizing patterns, understanding language, making predictions, interpreting images, planning actions, solving problems and generating content.
You already encounter AI in ordinary digital services. Search engines can interpret queries and rank results; email systems can identify unwanted messages; navigation systems can estimate routes and travel times; recommendation systems can predict content you may find useful; and generative AI systems can produce text, images, code and other content.
What Does Artificial Intelligence Mean?
There is no single technology called AI. It is an umbrella term covering many techniques and applications. Some AI systems use machine learning, some use neural networks, and many real products combine learned models with conventional software.
A Simple Analogy
Imagine teaching someone to identify fruit. One approach is to write explicit rules such as “if it is yellow and curved, it may be a banana.” That resembles traditional rule-based programming. Another approach is to provide many examples and allow a learning process to discover useful patterns. That idea is central to machine learning.
Modern applications often combine both approaches. A learned model can perform a difficult prediction while ordinary software handles the surrounding application logic.
Traditional Software vs AI
Traditional software is often created by explicitly writing rules and instructions. A simple calculator, for example, can contain formulas that specify exactly how an input should be transformed into an output.
Many machine-learning systems work differently. Instead of manually specifying every possible rule, developers provide data and an algorithm produces a model that captures patterns in that data. The distinction is useful, but it is not absolute: real-world software frequently contains both conventional programming and AI components.
Examples of AI You May Already Use
- Search: interpreting queries and ranking results.
- Recommendations: predicting content or products that may interest you.
- Spam filtering: identifying patterns associated with unwanted messages.
- Speech recognition: converting spoken language into text.
- Computer vision: detecting and classifying patterns in images.
- Navigation: estimating routes and travel times.
- Generative AI: producing text, images, audio, video or code from instructions.
AI Is an Umbrella Term
AI is not synonymous with ChatGPT, robots, machine learning or generative AI. Think of AI as the broad field, with machine learning as a major approach within it. Deep learning is a major family of machine-learning techniques based heavily on multi-layer neural networks. Generative AI describes systems designed to generate new content. Large language models are an important class of models used for language-related tasks.
Narrow AI and General AI
Most AI systems used today are designed for particular tasks or classes of tasks. This is often called narrow AI. A model might be excellent at recognizing objects, generating text or recommending products while not possessing broad human-like ability across unrelated domains.
You may also hear artificial general intelligence (AGI), a term generally used for a hypothetical level of machine intelligence with broad and flexible capabilities across many intellectual tasks. Definitions vary, and AGI should not simply be treated as another name for today's ordinary AI applications.
AI Does Not Automatically Mean Human Understanding
A system can produce remarkably useful results without understanding the world exactly as a person does. AI systems can make errors, reflect problems in data, misinterpret instructions, or produce outputs that sound convincing but are incorrect. Understanding both capability and limitation is a central part of AI literacy.
AI, Data and Models
Many modern AI systems depend on data. During development, data can be used to train a model to identify patterns. Once trained, the model can process new inputs and produce outputs.
For example, an image model can learn statistical patterns from a large collection of images, while a language model can learn patterns in language from large datasets. A trained model represents learned statistical relationships in a mathematical form. The use of a trained model to produce a result is commonly called inference.
Generative AI
Generative AI has attracted attention because it can create new content in response to instructions. Depending on the system, that can include text, code, images, audio or video.
A language model does not simply retrieve a stored paragraph from a conventional database. It generates an output based on patterns learned during training and the information supplied in the current context. Fluent wording therefore does not guarantee factual accuracy.
Common Misconceptions
AI is always a robot
Robots are physical machines. AI is a field of computing. An AI system can exist entirely as software.
AI always learns while you use it
Many deployed systems do not continuously retrain themselves from every conversation. Training and deployment are separate processes, although some systems may have mechanisms for updates or new information.
AI always gives the correct answer
AI can be extremely useful and still be wrong. Accuracy depends on the system, task, data, context and other factors.
AI is one single technology
AI includes many techniques, models, architectures and applications. There is no single algorithm that represents all of AI.
Why Learn AI?
AI literacy is useful even if you never become an AI engineer. It can help you evaluate AI-generated information, use AI tools effectively, identify suitable use cases, understand limitations, and make better decisions about privacy and security. For developers, the subject extends into models, APIs, data, retrieval, automation and evaluation.
Putting It All Together
AI is a broad field focused on computational systems capable of tasks associated with intelligence. Machine learning is one of the most important approaches in modern AI. Deep learning is a major family of machine-learning techniques. Generative AI describes systems that create content, and large language models are an important class of language models.
Learning how these terms relate gives you the vocabulary needed for the rest of this course.