Artificial Intelligence (AI)
AI is the broad field of building systems that perform tasks associated with human intelligence — reasoning, recognizing patterns, making decisions, and generating content.
What Is AI?
Artificial intelligence is the broad field of building systems that perform tasks normally associated with human intelligence: reasoning about a situation, recognizing patterns in data, making a decision, or generating new content.
That's intentionally a wide definition, because AI is a field, not a single technique. A rules-based chess engine from the 1990s, a spam filter from the 2000s, and a modern language model are all "AI" — they just use very different methods to get there.
Key Idea
AI is the goal (build systems that act intelligently). Machine learning and deep learning are specific approaches to reaching that goal — not the only ones, and not synonyms for it.
People often use "narrow AI" and "general AI" to separate two very different ambitions. Narrow AI is built to do one thing well — recognize faces, translate text, recommend a product. Every AI system in production use today is narrow AI. "General AI" — a system with human-like flexibility across any task — remains a research goal, not something that exists yet.
A Mental Model
A useful (if imperfect) way to place related terms you'll run into is as a chain of increasingly specific approaches:
Classical ML
Traditional approachStatistical methods like regression and decision trees.
Symbolic AI
Alternative approachRule-based reasoning and logic, without learning from data.
Warning
This is a conceptual map for orientation, not a strict taxonomy. Plenty of real AI systems — a rules engine, a search algorithm, a classical statistical model — sit under "AI" without ever touching a neural network. Not every AI system uses machine learning, and not every machine learning system is deep learning.
Where AI Shows Up
Most people interact with several of these before breakfast without thinking of them as "AI":
- Recommendation systems — suggesting a show, a song, or a product based on patterns in behavior.
- Fraud detection — flagging a transaction that looks statistically unusual for an account.
- Speech recognition — turning spoken audio into text.
- Computer vision — detecting objects, faces, or defects in images.
- Generative AI — producing new text, images, or audio rather than just classifying existing data.
Common Mistakes
"AI" meaning "neural network" or "LLM"
A lot of AI in production — fraud rules, recommendation heuristics, classical statistical models — has nothing to do with neural networks or language models.
Treating AI and AGI as the same thing
Every deployed AI system today is narrow AI, built for a specific task. General-purpose, human-level flexibility is a research goal, not a shipped product.
Assuming an AI system reasons the way a person does
A system can produce useful, intelligent-looking output through pattern matching and statistics without "understanding" the task the way a human would.
Interview Question
What is artificial intelligence?
AI is the field of building systems that perform tasks associated with human intelligence — things like recognizing patterns, making decisions, or generating content. It's a broad field, not one technique: it includes everything from rule-based systems to statistical machine learning to deep neural networks. In practice, almost all AI in production today is narrow AI, built to do one task well, rather than general-purpose intelligence.
What an interviewer may ask next
- How is machine learning different from AI more broadly?
- What is the difference between narrow AI and general AI?
- Can you give an example of AI that is not machine learning?
Explain It in 30 Seconds
AI is the broad field of building systems that do things we associate with human intelligence — reasoning, recognizing patterns, making decisions, generating content. Machine learning and deep learning are specific approaches within that field, not the whole thing. Almost everything deployed today is narrow AI: built for a specific task, not general-purpose intelligence.