Exam practice

Daraz and HamroBazar (others: Foodmandu, Pathao for food delivery).

E-governance is the use of digital devices, the internet and communication technology by the government to provide services and information to citizens, businesses and its own offices, such as online passport applications, tax filing and results, making government faster, more transparent and easier to reach.

• Natural language processing: voice assistants and translation apps that understand and answer in human language.
• Computer vision: face unlock on phones, reading number plates.

(Others: gaming, automated grading, personalized learning, expert systems, handwriting recognition, intelligent robots.)

ChatGPT (by OpenAI) and Gemini (by Google). (Microsoft’s Copilot can too.)

• Education: exploring the inside of the body or the Solar System in a virtual classroom.
• Training: pilots practicing in flight simulators, surgeons practicing operations.

(Others: games, health, business, tourism, military, aeronautics.)

• Students can learn anytime and anywhere, at their own pace.
• It gives access to a wide variety of learning materials, often at lower cost.

• The Nagarik App, which gives citizens many government services and documents on their phone.
• Applying for a passport (or a driving license) online.

(Others: filing tax online with the Inland Revenue Department, SEE results online, online company registration.)

• The product delivered may be different from what was shown online.
• There is a risk of online scams and fraud (and delivery can be late).

• Storing and sharing files online (Google Drive, OneDrive), to open them from any device.
• Using software online without installing it (Gmail, Google Docs), or renting servers to run a website.

ChatGPT, Gemini, Copilot and DeepSeek (others: Claude; Midjourney and DALL·E for pictures).

Artificial intelligence is the branch of computer science that makes machines do things that need intelligence when people do them: learning from experience, reasoning, solving problems, understanding language and recognizing pictures.
Five applications in daily life:

• i.Voice assistants and translation (natural language processing): asking a phone a question by voice, or translating a sign from English to Nepali.
• ii.Face unlock and photo search (computer vision): the phone recognizes its owner’s face; a photo app finds all the pictures of mountains.
• iii.Maps and navigation: routes chosen from live traffic, and travel times predicted.
• iv.Spam filters and fraud alerts: unwanted email kept out of the inbox; a bank blocking an unusual payment.
• v.Personalized learning and recommendations: learning apps that give more practice where you make mistakes; video apps that suggest what to watch.

(Also: chatbots, automated grading, games, robots, expert systems in hospitals.)

Machine learning is the part of AI in which a computer learns from data, finding patterns in examples and getting better with experience, instead of following step-by-step rules written by a programmer. To recognize cats and dogs, it is given thousands of labeled photos, not rules about ears and tails.
Its relation to AI: AI is the broad field of making machines smart; ML is a subset of it, one way of making a machine smart, and the way most AI works today. All ML is AI, but not all AI is ML: a chess program following rules written by people is AI without learning.
Examples:

• A map app (AI) predicts journey times by machine learning from millions of past journeys.
• A spam filter learns from emails people marked as spam.
• A voice assistant (AI) understands speech with a model that learned from thousands of hours of recorded speech.

• i.Education: learning apps learn which questions each student gets wrong and give more practice on those topics at the right level; quizzes are marked automatically.
• ii.Healthcare: models trained on thousands of X-rays and eye scans help doctors find disease (tuberculosis, damage to the eye from diabetes) early, and predict which patients need care first.
• iii.Transportation: map apps predict traffic and choose faster routes; self-driving cars recognize roads, signs, people and vehicles; ride apps predict demand and price.
• iv.Banking: fraud detection notices unusual payments and stops online theft.

(Also: spam filtering, speech recognition, recommendations, cybersecurity.)

How each works:

• Supervised: a model is shown thousands of photos of tigers and lions, each marked with its name. It learns which features go with each name. Given a new, unmarked photo, it answers “tiger” or “lion”.
• Unsupervised: the same photos are given without names. The model puts similar photos together and ends with two groups, striped animals and maned animals, but it can’t name them; a person looks at the groups and names them.

Robots sense their surroundings, decide and act; they work tirelessly, precisely and safely in places dangerous for people, so industries use them for dangerous, difficult and dull work.

• Manufacturing: robot arms weld, paint and assemble cars and electronics, day and night, with the same precision every time; warehouse robots carry goods to packers. Production is faster and cheaper, and workers are kept away from heat, fumes and heavy loads, though some routine jobs disappear and new jobs appear in programming and repairing robots.
• Healthcare: surgical robots, guided by a surgeon, operate through small cuts, so patients heal faster; robots carry medicines in hospitals, disinfect rooms, and help patients relearn to walk.
• Education: robot kits teach students programming, sensors and engineering (robotics clubs and competitions in Nepal’s schools); robots help children with special needs practice.
• Other sectors: robots defuse bombs, explore the deep sea and Mars, fight fires, and work in agriculture (picking fruit, spraying crops) and homes (vacuum cleaners).

Generative AI is AI that creates new content (text, pictures, music, video and computer code) after learning from huge amounts of examples. A language model such as ChatGPT writes by predicting, again and again, the most likely next word.
Importance:

• It makes a first draft in seconds: an email, a story, a summary, a program.
• It helps people who can’t draw, design or code to make things.
• It helps students understand: explaining a topic in simple words, or in Nepali.

How it is changing content:

• Text: drafting, rewriting, translating and summarizing in seconds; writers edit rather than start from nothing.
• Images: pictures made from a sentence for posters, books and advertisements.
• Video and sound: videos, voices and music generated from text.
• Code: assistants such as Copilot suggest whole functions.

Concerns: it can be confidently wrong; it can copy bias from its data; deepfakes (fake videos or voices of real people) can spread lies; questions of copyright and of honesty in schoolwork. So its work must be checked, and labeled as AI-made.

In a smart city, thousands of IoT devices (sensors, cameras, meters) collect data about traffic, energy, water, air and waste and share it over the internet, so the city can be managed by facts, often automatically.

• Traffic: sensors and cameras count vehicles, and traffic lights change their timing with the traffic; apps show buses’ positions and free parking spaces, so there are fewer jams.
• Energy: streetlights dim when nobody is near and brighten when someone comes; smart meters show homes how much electricity they use.
• Waste: bins report when they are full, so trucks collect only full ones, saving fuel.
• Water: sensors find leaks in pipes, and smart meters measure use.
• Safety and environment: air-quality sensors warn people on badly polluted days; smoke detectors call the fire brigade; river sensors warn of floods.

Problems to solve: the cost, keeping the devices secure from hackers, and the privacy of people being watched.

E-commerce is buying and selling goods and services over the internet, as on Daraz or Foodmandu.
Merits:

• Convenient: shop from home at any hour (ordering a phone on Daraz at midnight).
• Saves time and travel: no going from shop to shop.
• More choice and easy comparison: many sellers’ prices and reviews side by side; products from other countries.
• Home delivery, and digital payment with wallets such as eSewa or Khalti.
• For sellers: a small business can reach customers all over Nepal without a shop.

Demerits:

• What arrives may differ from the photo (wrong size, poor quality), and returns can be difficult.
• Scams and fraud: fake sellers, fake payment pages, stolen card details.
• Late or failed delivery, especially outside the cities.
• Less social contact, and dependence on technology and the internet.
• Small local shops can lose customers.

On balance, e-commerce is useful when we buy from trusted sellers, check reviews, prefer cash on delivery from new sellers, and keep payment details safe.

E-governance puts government services online, so citizens get them faster, more fairly and without traveling to offices.

• Faster, less paperwork: forms are filled in online and tracked; documents are stored digitally (the Nagarik App keeps a citizen’s documents on their phone).
• Transparency and less corruption: fees are paid online and each application’s progress can be checked, so there is less need, and less chance, for bribes.
• Equal access: a person in a remote district can use the same service as someone in Kathmandu.

Examples:

• Transport: applying online for a driving license and booking a date for the test, instead of queuing all night.
• Education: SEE and other results published online; scholarships and teachers’ licenses applied for online.
• Health: online registration and appointments at hospitals; health insurance records kept digitally.
• Others: passport applications online, paying taxes and electricity bills online.

To reach everyone, the government must also close the digital divide and teach digital skills.

Advantages:

• Learning anytime, anywhere, at the student’s own pace; a lesson missed can be watched later.
• A wide variety of materials: videos, simulations, quizzes, books and lessons from the best teachers anywhere.
• Often cheaper: no travel, free resources.
• Flexible for students who work, live far away or are ill.

Changes for students: learning is no longer limited to the classroom, the teacher and the textbook; students search, watch, practice with instant feedback and take responsibility for their own learning; they also need self-discipline and good internet.

Changes for teachers: teachers use online classes, share materials and set and mark assignments online (Google Classroom); they become guides who help students learn, not only the source of knowledge, and need digital skills.

But: less face-to-face contact, technical problems, the digital divide in Nepal, and content of mixed quality. The best results often come from blended learning: classroom teaching supported by e-education.

Cloud computing is the use of computing services (storage, servers, software, databases) over the internet from a provider’s data centers, paid for as used, instead of owning and running them yourself. Providers include Amazon Web Services, Microsoft Azure and Google Cloud.
Its main services:

• i.SaaS (Software as a Service): complete applications used over the internet with nothing installed; the provider runs everything. Examples: Gmail, Google Docs, Microsoft 365, Google Drive.
• ii.PaaS (Platform as a Service): a ready platform (operating system, programming languages, database, tools) on which programmers build, test and run their own programs without managing servers. Examples: Google App Engine, Azure App Service.
• iii.IaaS (Infrastructure as a Service): basic hardware rented over the internet (virtual servers, storage, networks); the customer installs and manages the operating system and software. Examples: Amazon EC2, Amazon S3, Azure Virtual Machines.
• iv.STaaS (Storage as a Service): space to keep files and backups online, paid for by how much is used, copied in several places so a broken disk loses nothing. Examples: Amazon S3 for businesses; for people, the storage behind Google Drive, Dropbox or iCloud. (Some lists give FaaS, Function as a Service, instead: a programmer uploads one small function and the provider runs it only when it is called, as AWS Lambda does.)