• Foundation
  • About 14 min
  • 5 practice questions

AI around you: phones, maps, shopping, banking and voice assistants

Explain in plain words what artificial intelligence is, how it learns from examples, and spot where it already works in your phone, maps, shopping apps, bank alerts and translation.

If something looks wrong, please tell us.

You probably used artificial intelligence (AI) several times today without noticing. When your phone unlocks by looking at your face, when a map warns you of a traffic jam, or when your bank sends an alert about a payment you did not make, AI is at work. Competitive exams ask about AI in general-studies and science-and-technology papers, and the questions are much easier once you see AI as something ordinary and practical rather than as science fiction.

What AI actually is

Artificial intelligence is the ability of a computer system to do tasks we normally link with human intelligence: recognising faces and objects, understanding speech, translating languages, making predictions and taking decisions. Most AI you meet today is machine learning: instead of a programmer writing every rule, the system is shown a very large number of examples and finds the patterns in them by itself. A system shown thousands of labelled photos of cats and dogs learns to tell a new photo of a cat from a dog.

Today's AI is narrow AI: each system is good at one kind of task. The model that recognises your face cannot drive a car or write an essay. The idea of a machine that can do any intellectual task a person can do is called general AI, and it does not exist yet.

How an AI system learns

How an AI system learns

  1. Collect data
  2. Train
  3. Test
  4. Use
  5. Improve

Then back to step 1: Collect data

Feedback and new data are used to train it again.
  1. Collect data: gather many examples, such as photos, voice recordings, past transactions or traffic speeds.
  2. Train: the system adjusts itself until it can match the examples correctly, for instance saying 'cat' for cat photos.
  3. Test: check it on new examples it has never seen, to find out how often it is right.
  4. Use: put it to work in an app or service.
  5. Improve: collect feedback and new data, and train it again.

AI in your daily life

Common everyday uses of AI
Everyday useWhat the AI doesWhat it learns from
Face unlock on a phoneRecognises whether the face in front of the camera matches the owner'sThe owner's face, recorded when face unlock is set up
Phone keyboardSuggests the next word and corrects spellingPatterns in large amounts of text, and your own typing
Photo gallery searchFinds photos of 'dog', 'beach' or 'birthday cake' without labels you typedMillions of labelled images
Maps and trafficEstimates travel time and suggests a faster routeSpeeds reported by many phones on the road and past traffic patterns
Shopping and video appsRecommends products or videos you may likeWhat you and people with similar tastes bought or watched
Bank fraud alertsFlags a payment that looks unusual for youYour normal pattern of spending
Voice assistantsTurns speech in Hindi, Tamil, Bengali or other languages into text, works out the request and repliesRecordings of speech matched with written text
Translation appsTranslates text or speech between languagesLarge collections of the same sentences in two languages

Worked example 1: a fraud alert

Ramesh lives in Lucknow and usually makes small UPI payments during the day. At 2 a.m. a large payment is attempted from a new device in another city. What does the bank's AI do, and why? Working: the system has learned Ramesh's usual pattern - amounts, times, places and devices. This payment differs on all four, so the system scores it as risky and blocks it or asks Ramesh to confirm. Note what the AI does not do: it does not 'know' that a crime is happening. It only notices that the payment does not fit the pattern. That is why a genuine but unusual payment, such as paying a hospital bill while travelling, can also trigger an alert.

How the bank's AI handles Ramesh's 2 a.m. payment

  1. Learns his usual pattern Amounts, times, places and devices.
  2. Payment differs on all four
  3. Scores it as risky
  4. Blocks it or asks him to confirm

Worked example 2: why the app keeps showing cooking videos

Meena watches several Tamil cooking videos on a video app. Next day her home screen is full of Tamil cooking videos. Working: the recommendation system has learned that people who watched these videos often watched similar ones, so it predicts she will too. This is useful, but it can trap a viewer inside a narrow set of topics, sometimes called a filter bubble. Meena can widen her feed by searching for new topics or clearing her watch history.

Limits to remember

AI can be wrong, and it is only as good as the data it learned from. If the training data left some groups out, the system may work worse for them; this is called bias. Many AI services also depend on personal data, so privacy matters: check what permissions an app asks for.

AI finds patterns in data and uses them to recognise, predict or recommend. It is only as good as the data it learned from, and it can be wrong.

Key takeaways: AI is a computer system doing tasks linked with human intelligence. Machine learning lets it learn from examples instead of fixed rules. Face unlock, keyboards, photo search, maps, recommendations, fraud alerts, voice assistants and translation are all everyday AI. Today's AI is narrow, can make mistakes and can carry bias from its data.

Check your understanding

  1. 1. Which statement best describes machine learning?

    1. A programmer writes every rule the computer must follow
    2. The system finds patterns from many examples instead of being given every rule
    3. A robot that looks and moves like a human
    4. A faster internet connection for apps
    Show answer

    Answer: B. Machine learning means the system learns patterns from a large number of examples (data). It is not hand-written rules, it need not involve a robot, and it has nothing to do with internet speed.

  2. 2. A map app tells you a road is jammed even though no one typed in a traffic report. What is it mainly learning from?

    1. Newspaper reports about traffic
    2. The number of petrol pumps on the route
    3. Weather satellites only
    4. Speeds reported by many phones on the road and past traffic patterns
    Show answer

    Answer: D. Map apps estimate traffic from the movement speeds of many phones on the road, combined with past patterns for that time and place. Newspapers, weather satellites or petrol pumps do not give live road speeds.

  3. 3. Your bank blocks a large late-night payment from a new device and asks you to confirm it. The AI did this because the payment:

    1. Did not fit your usual pattern of amount, time, place and device
    2. Was made using UPI, which the AI always blocks
    3. Was below the minimum amount
    4. Came from your regular phone
    Show answer

    Answer: A. Fraud-detection AI learns each customer's normal pattern and flags payments that differ from it. UPI payments are not blocked as a rule, the amount was large rather than small, and the device was new, not your regular phone.

  4. 4. Which of these is an example of narrow AI?

    1. A machine that can do any intellectual task a human can do
    2. A person using a calculator
    3. A system that only recognises faces to unlock a phone
    4. A television set
    Show answer

    Answer: C. Narrow AI is good at one kind of task, like face recognition. A machine able to do any intellectual task would be general AI, which does not exist yet. A person with a calculator and an ordinary television are not AI systems.

  5. 5. A face-recognition system works well for some people but often fails for people from groups that were rare in its training photos. This problem is called:

    1. Encryption
    2. Bias
    3. Bandwidth
    4. Compression
    Show answer

    Answer: B. When training data leaves some groups out or under-represents them, the system works worse for those groups. This is bias. Encryption, bandwidth and compression are unrelated computing terms.

Privacy · Terms · Refunds · Delivery