AI for Beginners: Understand Artificial Intelligence from Zero

Course Introduction

Welcome to the world of Artificial Intelligence. This course takes you from zero to a working understanding of AI — no technical background required. We explain what AI is, how it learns, and how it is already changing businesses like yours, whether you run a pet shop, a wholesale business, or an online store.

Learning objectives: by the end of this course you will be able to (1) explain AI in plain language, (2) tell the difference between the main types of AI, (3) identify practical AI applications for your business, (4) recognise AI's limitations and common myths, and (5) build a concrete 30-day action plan.

1.1 What Is AI? Definition and History

Artificial Intelligence (AI) is technology that lets computers perform tasks that normally require human intelligence — recognising patterns, understanding language, making predictions, and generating content. Not magic, not consciousness: at its core, AI finds statistical patterns in data and applies them to new situations.

Three milestones you should know

• 1956 — AI is born. The term "Artificial Intelligence" is coined at the Dartmouth Conference. Early decades see waves of optimism and funding freezes ("AI winters").
• 2012 — the deep learning breakthrough. A neural network crushes the ImageNet image-recognition competition, kick-starting today's AI boom.
• 2022 — generative AI goes mainstream. ChatGPT reaches 100 million users in two months, the fastest-adopted consumer product in history at the time.

How to start understanding AI in 3 steps

1. Find one AI application in your industry. Open ChatGPT (chat.openai.com), register free, and ask: "I run a [your industry] business — what problems can AI solve for me?" Note the three suggestions you understand best.
2. Separate "media AI" from "practical AI". News covers frontier research (robots, self-driving). Businesses actually use mature tools: text generation, data analysis, customer-service automation. Beginners should start with the latter — fastest time to value.
3. Use the "data + algorithm = model" framework. For any AI product, ask three questions: What data was it trained on? What learning method does it use? What is the resulting model good at? This framework lets you evaluate 90% of AI marketing claims.

1.2 Types of AI: Narrow vs General vs Super

Everything you can use today is narrow AI: systems trained for one task family, however impressive. A chess AI cannot write emails; ChatGPT cannot drive a car. General AI (human-level flexibility across all tasks) and super AI remain research concepts, not products. This matters practically: you should match each tool to the specific task it was built for, and be sceptical of any vendor promising "one AI for everything."

Choosing tools by capability type

1.3 Machine Learning vs Deep Learning vs Generative AI

These three terms describe nested layers: generative AI is a subset of deep learning, which is a subset of machine learning, which is a subset of AI. Machine learning is the umbrella term for any method where computers learn patterns from examples instead of following hand-written rules. Deep learning is the branch that uses many-layered neural networks — the "deep" refers to layers. Generative AI is the application layer that creates new content: text, images, audio, video.

1.4 How AI "Learns": Data, Models, Algorithms

The core formula of machine learning is data + algorithm = model. Data is the training material — sales records, customer interactions, images. The algorithm is the learning method — how the system extracts patterns. The model is the finished product that can make predictions on new data. Data quality sets the ceiling on performance: the industry saying is "garbage in, garbage out."

Hands-on: experience how AI "learns"

1. Prepare a dataset. In a spreadsheet, list eight weeks of two columns: promotion spend and store revenue.
2. Ask AI to find the pattern. Paste the data into ChatGPT (or use Excel's built-in analysis) and type: "Analyse the relationship between these two columns. If I spend $5,000 more on promotions next week, what revenue should I expect?"
3. Observe the answer. The AI finds an approximate relationship (that's the "model") and gives a prediction with a confidence range.
4. Understand garbage-in-garbage-out. Delete a few cells and ask again — the prediction becomes unstable. That's data quality directly shaping AI performance.

1.5 Neural Network Basics

A neural network is a stack of simple decision-making units organised in layers, inspired loosely by the brain. Think of a company: frontline staff each receive information (input layer), middle managers digest and combine it (hidden layers), and leadership makes the final call (output layer). The "depth" in deep learning is simply the number of middle layers. Training works by measuring how wrong the final answer is, then adjusting every layer's judgment criteria a little — big corrections for big mistakes, small tweaks for small ones (called backpropagation).

1.6 AI in Everyday Life — and Your Business

You already use AI daily: Google Maps rerouting (traffic prediction), Netflix recommendations (collaborative filtering), phone photo grouping (computer vision), keyboard next-word suggestions (language models). Business AI is the same set of capabilities pointed at your own data. The practical skill is "transfer thinking": spotting which consumer AI experience maps onto a business problem you have.

Transfer exercise

1. List three AI features you use daily and what problem each solves (e.g. Maps avoids jams = real-time traffic prediction).
2. Transfer it. Ask: "Does my business have a similar prediction need?" (Which products will run out next week? Which customers are about to churn?)
3. Pick one need and try a first version in ChatGPT with your real (de-sensitised) data.

1.7 Global AI Trends That Matter to SMBs

For small businesses, the winning strategy is to be an "application-layer follower": don't chase model technology — let big-vendor competition make tools cheaper and easier, then adopt. Your competitive advantage comes from industry knowledge applied through AI, not from AI itself.

1.8 AI Limitations, Bias and Myths

AI systems can produce confident, fluent, entirely wrong answers — the industry calls this hallucination. Fluency is not accuracy. Models also inherit biases from training data, which matters most in hiring, lending and any people-decision context. The professional habit: treat every factual claim from AI as a draft that needs a source check.

Hands-on: witness a hallucination safely

1. Bait the AI. Ask ChatGPT an obscure question with no real answer (for example, "What was the pass rate of the 2024 Hong Kong pet-grooming licence exam?" — it doesn't exist).
2. Watch the hallucination. The AI will likely invent a plausible-sounding number.
3. Learn the defence. Add "If you are not sure, say so directly" to your prompt — hallucination rates drop noticeably.
4. Build the habit. Every fact, figure and legal reference from AI gets verified against an official source before it leaves your desk.

1.9 How AI Reshapes Work and the Economy

AI replaces tasks, not jobs. When ATMs arrived, bank teller numbers actually rose for years — branches got cheaper to open, and teller work shifted from counting cash to sales and advisory. The same reorganisation is underway: routine tasks automate, while judgment, creativity and human trust gain value.

1.10 Course Summary + Quiz