What is AI?

When you ask Siri what time it is, when Google translates a sign in Karachi from English to Urdu, when YouTube guesses the next song you might like, when a bank decides whether to approve a loan, you are watching a kind of system at work. People call this system artificial intelligence. The phrase is everywhere. The phrase is also imprecise. In this lesson we are going to be specific about what AI actually is, what it is not, and why the difference matters for the work you will do as a Pakistani learner growing up alongside this technology.

Way one to think about AI: a system that learns patterns from many examples. A spam filter does not have a list of every possible spam message. It has seen millions of emails labelled spam and not-spam, and it has worked out, over time, which words and shapes tend to mean spam. When a new message arrives, the filter does not know for sure. It guesses, with a confidence number, based on what it has seen before. AI in this sense is pattern recognition at a scale a human reviewer could never match.

Way two to think about AI: a system that produces outputs from inputs by following a learned function, not a rule book a human wrote. A traditional program for converting Celsius to Fahrenheit follows the rule the programmer typed: multiply by 9, divide by 5, add 32. An AI translation program does not follow a rule book of grammar. It has been trained on millions of sentence pairs in two languages and has worked out, internally, a function that maps one to the other. Nobody, not even the people who built it, can read the function as if it were a recipe. The system runs but its inner workings are partly opaque.

What AI is not, in 2026: it is not a thinking being, it is not conscious, and it is not a friend. A model trained to converse can produce sentences that sound warm, sincere, even insightful. The sentences are real. The warmth is statistical. We say this not to shame anyone who has felt comforted by an AI conversation, because the comfort is also real, but to draw a line we will return to throughout this course. The line matters when the conversation turns to something serious: a medical question, an exam result, a family difficulty, a moment of distress. At that point you need a person who can be accountable to you over time. AI is a tool, used well, useful. It is not a substitute for relationship and it is not a substitute for expertise.

Why this matters in Pakistan, specifically. AI literacy is already shaping who gets hired, who gets a loan, who gets a visa, who gets admitted to which programme. These decisions are not made on whether the person, in their own life, can recite a definition. They are made on whether the person can use AI tools wisely, recognise when an AI system is producing confident nonsense, and explain the difference to someone else. The HEC's mandate that every undergraduate course include three credits of AI starting Fall 2026 is a recognition of this. Your job in this course is not to memorise definitions. Your job is to build the kind of practical, judgement-loaded understanding that will travel with you into work and into the rest of your life.