Chapter 1 · Information Technology and Society
1-3 AI in Daily Life and Industry
AI is already part of many everyday services. A video site recommends the next clip, a voice assistant answers a spoken question, a translation app converts text on a foreign menu, and a phone unlocks by recognising your face. In industry, AI analyses medical images, predicts when a machine will fail, and plans delivery routes. It is effective at identifying patterns in data — but it is not suited to every task.
Guiding question
Where is AI used in daily life and industry, what is it good at, and what needs caution?
Learning outcome. Give examples of how AI is used in daily life and various industries, and explain what AI is good at and what requires caution when using it.
With a partner. List three services you used this week that you think used AI — a video recommendation, a voice assistant, a translation. For each, predict what data the AI needed to do its job. Then decide together: which one would matter most if it made a mistake?
Use of AI in daily life
| Service | Role of AI | Examples |
|---|---|---|
| Recommendation systems | Predict preferences from past behaviour data and display recommendations. | YouTube, Amazon, Spotify, etc. |
| Voice assistants | Recognise a voice, understand commands, and execute them. | Siri, Google Assistant, etc. |
| Machine translation | Automatically translate text into different languages. | Google Translate, DeepL, etc. |
| Face recognition | Automatically detect and identify human faces in photographs. | Smartphone unlocking, etc. |
Use of AI in industry
| Industry | Examples of AI use |
|---|---|
| Healthcare | Image-diagnosis AI (detecting diseases from X-ray and CT images), drug-discovery support |
| Agriculture | Predicting harvest timing, detecting pests and diseases |
| Manufacturing | Automation of product quality inspection, predictive maintenance |
| Logistics | Optimization of delivery routes |
Characteristics of AI, and cautions for use
What AI is good at
- • Finding and classifying patterns in complex data such as images and text
- • Recognition and generation of images, audio and text
- • Probabilistic reasoning and prediction based on data
What requires caution when using AI
- • Ethical judgments (may lead to discrimination or prejudice)
- • Judgments involving personal information or privacy
- • Final decision-making (who is responsible for the result)
- • Accuracy of judgments when training data is biased
- • Generating content that differs from the facts (hallucination)
- • Issues of rights when using copyrighted works as training data
- • Loss of clarity about how a judgment was made (the black-box problem)
“Needs caution” does not mean “AI cannot be used”.
Hospitals do use image-diagnosis AI — it is in the industry table above. The caution is that a doctor confirms the result. The right conclusion is not “keep AI out of medicine” but “a human stays responsible for the final decision”. That distinction is worth being careful about, because a student who learns the first version has learned something untrue from a true list.
Everything in the “needs caution” column is a question about what SHOULD BE — what is fair, what is private, who is to blame, what we ought to do. Those answers are not in the data at any quantity, because they are not facts about the past. No amount of extra examples can produce them.
If you can hold that one sentence, you can rebuild both of the book’s lists in the exam without having memorised either.
The book’s wording, which is what an exam wants: recommendation system — a system that predicts preferences from past behaviour data and displays recommendations. (KEY TERMS, p.20)
Step 1 of 6 — All the site actually keeps is a grid of ticks
It does not know you. It does not store an opinion about you. It stores one row per person, with a tick for every video that person watched. Your row says you watched the football highlights and the physics revision, and nothing else. That grid is the entire input.
The grid is invented for teaching — the book gives no figures, and a real service compares millions of viewers rather than four. What is real is the method: a record of who watched what, a count of who resembles you, and a sum. Keep the book’s own wording for the exam — recommendation systems suggest content based on past behaviour data — and keep this picture for knowing what “based on” is doing in that sentence.
🧮 Working the same recommendation out by hand
- 1
Write down who watched what, as 1 for watched and 0 for not, across the five videos — football, cooking, physics, cats, guitar. You: 1,0,1,0,0 · Mona: 1,0,1,1,0 · Ali: 1,1,0,0,1 · Sara: 0,1,0,1,1 · Youssef: 1,0,1,0,1
Why this step: Same first move as every other example in this course: turn the situation into numbers, because that is the only thing the machine can work with. Each person becomes a row of five 1s and 0s, and this grid is the whole of what the site stores about anybody.
- 2
Count how many videos you and each other person BOTH watched. Mona: football and physics → 2. Youssef: football and physics → 2. Ali: football only → 1. Sara: none → 0.
Why this step: This count is the entire meaning of “someone like you” inside the machine. It is deliberately crude, and that is the honest picture — no understanding of what any video is about enters into it anywhere.
Most people slip here: Count only the videos you BOTH watched. Two people who both skipped the same video have not thereby shown they are similar — almost nobody watches almost everything, so counting shared 0s would make every pair of people look alike.
- 3
Cross out the videos already in your own row: football and physics.
Why this step: A recommendation has to be something you have not seen. Those two are out of the running before any scoring starts, which leaves cooking, cats and guitar.
- 4
Score each remaining video by adding the counts of everyone who watched it. Cooking: Ali (1) + Sara (0) = 1. Cats: Mona (2) + Sara (0) = 2. Guitar: Ali (1) + Sara (0) + Youssef (2) = 3.
Why this step: Adding the counts means a person who matches you strongly pushes harder than one who barely matches, and Sara — who shares nothing with you — adds 0, so her taste has no influence at all. Every number on this line came from step 2 and nowhere else.
Most people slip here: Do not simply follow your single closest match. Mona matches you as closely as anyone and Mona watched the cats — the cats still lose.
- 5
Take the highest score: guitar, with 3.
Why this step: Two people who partly match you both watched the guitar lesson, and 1 + 2 beats Mona’s 2 on her own. That is why the answer is the guitar lesson and not the cats — the arithmetic gave an answer that guessing would not have.
Most people slip here: Nothing about the guitar lesson itself was consulted. The system does not know what it is, whether it is any good, or whether it suits you.
Where the comparison breaks down: A real assistant who noticed you were buying something unsuitable would say so. This one cannot, because it has no idea what anything is. That is why the book puts final decision-making in the caution column — the system can tell you what is popular with people like you, and that is a different question from what is good for you.
The same underlying technology, pointed at photographs of people, has repeatedly been found to be far less accurate on some groups of faces than others — because the photo collections it learned from contained far more of some faces than others. Same maths, same method. The difference is that one is a question about what is in an image, and the other becomes a question about how a person is treated. That is the rule at the top of this page, in a single pair of examples.
Both patterns are widely documented across published studies of deployed AI systems.
This is the skill the exam-style question asks for. For each task, decide — then read the book’s reason.
Sorting thousands of photographs into “cat” and “not cat”
Ask yourself: is this finding a pattern, or is it deciding something about a person?
How the real paper is built
The ministry sets half multiple choice and half written answer in every subject. The two halves are not the same skill. In the multiple-choice half you have to spot the wrong options quickly — and the wrong options are built out of the mistakes students really make. In the written half nobody gives you the answer to recognise; you have to produce it, in your own words, with a reason.
Practise both. Being good at one half and weak at the other costs you exactly half the paper.
- recommendation system
- — predicts preferences from past behaviour data.
- voice assistant
- — recognises speech, understands commands, executes them.
- machine translation
- — translates text between languages automatically.
- face recognition
- — detects and identifies faces in photographs.
- predictive maintenance
- — predicting failures before they happen.
- hallucination
- — AI generating content that differs from the facts.
- black-box problem
- — when it is unclear how an AI reached its judgment.
Worked Example
(1) Match each service (a–d) with the AI technology it uses: A Recommendation system, B Voice assistant, C Machine translation, D Face recognition.
a YouTube recommends videos that match the user’s preferences
b Speaking to a smartphone to check the weather
c Translating a foreign-language website into one’s native language
d A smartphone camera automatically detects a person’s face
(2) Mark each statement with ○ if it is true or × if it is false.
A AI is good at probabilistic reasoning and prediction based on data.
B AI is good at recognising images and audio.
C It is fine to leave ethical judgments entirely to AI.
D AI judgments can become inaccurate when training data is biased.
1 · Answer the following questions.
2(1) · Which does NOT belong to “what requires caution when using AI”?
Mark ○ for each one that DOES require caution, and × for the one that does not.
Ethical judgments
Finding and classifying patterns
Judgments involving personal information or privacy
Final decision-making
2(2) · For each AI usage example, which industry is it most closely related to?
Match each description with the choice that fits it.
Optimizing delivery routes
Detecting diseases from X-ray images
Detecting pests and diseases
Automating product quality inspection
Think as an Engineer — Investigate & Decide
You have just discussed agriculture in the note above. Now investigate a different industry and decide.
1 · Investigate.
Choose one industry from the lesson other than agriculture — healthcare, manufacturing or logistics. Find one real, documented example of AI being used in it, and record the source you used.2 · Weigh it up.
Note one benefit and one caution of using AI in that industry, using the ideas of what AI is good at and what needs caution.3 · Decide.
Should this industry use AI for this task on its own, or with a human checking the result? Recommend one approach and give two reasons.
In a New Context
Lesson Question — Answered
AI is used widely in daily life — in recommendation systems, voice assistants, machine translation and face recognition — and in industry, across fields such as healthcare, agriculture, manufacturing and logistics.
AI is good at finding and classifying patterns in data, recognising and generating images, audio and text, and making probabilistic predictions from data. However, it needs caution where ethical judgments, personal information and privacy, and final decision-making are involved, and where training data is biased. So AI is a powerful tool for pattern-based tasks, but human judgment is still needed for responsibility, ethics and fairness.
1 · Quiz yourself first.
Cover the Point! sections and name the four everyday services, the four industries, the three things AI is good at, and as many of the seven cautions as you can.
2(1) · Select all that AI excels at.
Mark ○ for each one AI excels at, and × for each one that does not belong.
Ethical judgments
Finding features in images and text and classifying them
Judgments involving personal information
Probabilistic reasoning and prediction based on data
Decision-making that bears responsibility for results
Recognition and generation of images, audio and text
2(2) · Which is NOT an appropriate description of services that use AI?
Mark each statement with ○ if it is true or × if it is false.
A recommendation system predicts preferences from past behaviour data and displays recommendations.
A voice assistant recognises a voice, understands commands, and executes them.
Machine translation is a system in which humans manually translate foreign languages.
Face recognition automatically detects and identifies human faces in photographs.
3 · Read the passage and fill in the blanks.
AI is widely used in our daily lives. For example, recommendations of videos that match one’s preferences on a video site come from a ( a ) system. Also, AI technology is used in ( b ), with which one can speak to a smartphone to operate it, and in ( c ), which converts foreign-language text into one’s native language.
3(2) · Choose all that require caution when using AI.
Mark ○ for each one that requires caution, and × for each one that does not.
High-speed processing of large amounts of data
Ethical judgments
Judgments involving personal information or privacy
Recognition of images and audio
Reflect: Which use of AI in this lesson surprised you most? Where in your own life do you think AI is making decisions about you?
Challenge: Choose one task where AI needs caution (ethics, privacy, or final decisions). Suggest one rule that would make using AI for that task safer.
Keep reading — this lesson marks itself done automatically.