What is AI?

Objective

By the end of this lesson, you will be able to explain artificial intelligence in practical terms and identify where it appears in real software products.

Prerequisites

You should be comfortable using software products and have a basic idea of how web or mobile apps take input, process it, and return output.

Plain-English Explanation

Artificial intelligence is software that performs tasks we usually associate with human judgment: recognizing patterns, making predictions, generating text, understanding images, ranking options, or choosing a next action.

AI is not magic. It is a way to build software that can handle messy inputs where writing exact rules would be too slow, too brittle, or impossible.

For example, a normal app might say: if the button is clicked, save the form. An AI feature might say: given this messy resume and this job description, estimate how well they match.

Key Terms

Artificial Intelligence

Simple meaning: Software that performs tasks that seem to require human-like judgment.

Real-world analogy: Think of a skilled assistant who can sort, summarize, classify, or recommend based on examples.

Technical meaning: A broad field of computing focused on systems that perceive, reason, learn, generate, or act in ways that usually require intelligence.

Example: GigSwipe could use AI to summarize a resume and recommend jobs that match the candidate's experience.

Why it matters: AI is the umbrella term. Machine learning, deep learning, LLMs, RAG, and agents all sit under it.

Model

Simple meaning: A program that has learned patterns from data.

Real-world analogy: A model is like a person who has reviewed many examples and can make a reasonable guess about a new one.

Technical meaning: A mathematical function with parameters learned during training, used to transform inputs into outputs.

Example: A model might read a support ticket and predict whether it is about billing, login, or product feedback.

Why it matters: When builders say they are using AI, they are usually sending input to a model and using the output in a product workflow.

Mental Model

Traditional software is rule-first:

Human writes rules -> Program follows rules -> Output

AI software is pattern-first:

Examples train model -> Model receives new input -> Output

You still write normal software around the AI. The AI part handles the judgment-heavy step.

Example

Imagine FurOrbit wants to help a pet owner understand a veterinary report.

Without AI, the team might write hundreds of rules for medical terms, sentence formats, and possible conditions. With AI, the product can ask a language model to summarize the report, define terms, and suggest questions to ask the vet.

The app still needs product logic: user accounts, file upload, privacy, display states, and safety copy. The model only handles the language understanding step.

Code Walkthrough

This pseudocode shows the shape of many AI features:

type ResumeMatchInput = {
  resumeText: string;
  jobDescription: string;
};

type ResumeMatchResult = {
  score: number;
  strengths: string[];
  gaps: string[];
};

async function scoreResumeMatch(input: ResumeMatchInput): Promise<ResumeMatchResult> {
  const prompt = buildResumeMatchPrompt(input);
  const modelOutput = await callLanguageModel(prompt);

  return validateResumeMatchResult(modelOutput);
}

The model does the fuzzy judgment. The application still builds the prompt, calls the model, validates the response, and shows it clearly.

Common Mistakes

• Treating AI as a complete product instead of one capability inside a product.
• Assuming AI is always correct because it sounds confident.
• Skipping validation, safety checks, and user experience design.
• Choosing AI when simple rules would be cheaper, faster, and more reliable.

Exercise

Pick one app you use weekly. Write down one feature that uses exact rules and one feature that could benefit from AI.

Use this format:

App:
Rules-based feature:
Possible AI feature:
Why AI helps:

Reflection

In one paragraph, explain the difference between "AI as a feature" and "AI as a product."

Further Reading

• OpenAI API documentation: Introduction
• Google ML Crash Course: Framing machine learning
• Khan Academy: Intro to Artificial Intelligence