Revision notes

Generative AI and Prompting: Practice

The key points, rules to remember, tables and diagrams from all 5 lessons, on one page for quick revision. To keep a copy, use your browser's Print and choose "Save as PDF".

Lesson 1

What generative AI is: models that create text, images, audio and code

Kinds of generative AI (tools named only as examples)
OutputWhat it makesWell-known examples
TextAnswers, summaries, letters, translationsChatbots such as ChatGPT, Gemini and Claude
ImagesPictures from a written descriptionImage generators such as DALL-E, Midjourney and Stable Diffusion
AudioSpeech in a chosen voice, musicText-to-speech and voice tools
CodePrograms, spreadsheet formulas, bug fixesCoding assistants such as GitHub Copilot

How an LLM builds an answer, one token at a time

  1. Take the text so far Your prompt plus everything written up to now
  2. Rate every possible next token How likely each word or piece of a word is to come next
  3. Pick one token Usually a likely one; temperature adds some randomness
  4. Add it to the text Then the loop repeats until the answer is complete

Then back to step 1: Take the text so far

LetterWordMeaning
GGenerativeIt produces new text rather than only classifying text
PPre-trainedIt is first trained on a huge general text collection, then adapted
TTransformerA neural-network design that uses 'attention' to weigh how the words in a passage relate to one another

GPT = Generative Pre-trained Transformer. An LLM writes by predicting the next token, again and again; it predicts likely text, it does not look up verified facts.

How a diffusion model makes a picture

  1. Start from pure random noise
  2. Remove a little noise Steered by your text prompt
  3. Repeat step by step Each step reverses one step of added noise learned in training
  4. An image appears
SystemCreates new content?Type
Email app moving a message to spamNo - it labels an existing emailPredictive
Chatbot drafting a leave applicationYes - new textGenerative
Phone unlocking when it sees your faceNo - it matches a facePredictive
App turning 'a poster for a science fair' into a pictureYes - new imageGenerative

Key takeaways: generative AI creates new content; predictive AI sorts or scores existing content. An LLM writes by repeatedly predicting the next token after pre-training on huge text collections, then fine-tuning and human feedback. GPT means Generative Pre-trained Transformer. Diffusion models make images by removing noise step by step, guided by the prompt.

Read the full lesson

Lesson 2

Writing good prompts: role, task, context, format and examples

A strong prompt

  • Role Who to act as
  • Task The exact job to do - must always be clear
  • Context Background, level and limits
  • Format The shape of the answer
  • Examples A sample of what you want
PartWhat it tells the AIExample
RoleWho to act asYou are a patient maths teacher for competitive-exam students.
TaskThe exact job to doExplain how to calculate compound interest.
ContextBackground, level and limitsI am a beginner and already understand simple interest.
FormatThe shape of the answerUse five numbered steps, then one worked example in rupees.
ExamplesA sample of what you wantWrite questions in the style of this one: ...
Weak promptWhat is missingStronger prompt
Tell me about GSTTask, level, formatExplain GST to a beginner in five short points, then give one simple example of a shop bill.
Write about riversScope and purposeMake a revision table of six peninsular rivers of India with columns River, Where it rises, Where it drains.
Make questionsTopic, number, type, answersMake 5 MCQs on the Fundamental Rights, four options each, answers with one-line reasons at the end.
Explain thisWhat 'this' is, levelExplain the paragraph below in simple English for a Class 8 student, in under 120 words: ...

Strong prompt = Role + Task + Context + Format (+ Examples). Be specific, ask for one job at a time, and refine with follow-up messages.

Refining an answer with follow-ups

  1. Explain inflation. Reply: 400 words with unfamiliar terms
  2. Shorten this to about 100 words.
  3. Now explain it in simple Hindi.
  4. Give it as a table: Term, Meaning, Everyday example.
  5. Ask me three questions to test whether I understood.
Each follow-up keeps the earlier conversation in view, so there is no need to start again.

Key takeaways: state the task clearly, then add role, context, format and examples as needed. Turn weak prompts into strong ones by naming the topic, audience, size and layout. Ask for step-by-step working so you can check it, and improve answers with short follow-up messages instead of retyping everything.

Read the full lesson

Lesson 3

Using AI to study for exams honestly: plans, explanations, practice and translation

Who does which job

  • AI chatbot: understand and practise
    • Revision plans
    • Simpler explanations
    • Practice questions
    • Translation of notes
    • Feedback on your answers
  • Official source or textbook: confirm facts
    • Exam dates, fees, eligibility, vacancies
    • Rules, laws, constitutional provisions
    • Current affairs
    • History, geography and science facts
    • Answer keys of practice questions
UseSample promptWhat you must still do
Revision planI have 30 days and 3 hours a day for reasoning, English and general studies; I am weak in English. Make a weekly plan.Adjust it to your real routine and follow it.
Simpler explanationExplain federalism in simple words with an everyday example.Explain it back in your own words.
Practice questionsMake 5 MCQs on the Indian monsoon, answers at the end.Check every answer key in your textbook.
TranslationTranslate this paragraph of my notes into simple Hindi.Check technical terms against a standard Hindi-medium book.
FeedbackHere is my answer on soil erosion. Point out gaps and unclear sentences.Rewrite it yourself.

The teach-back loop

  1. Ask for a simple explanation For a beginner, with an everyday example or analogy
  2. Close the chat and explain it yourself Aloud or on paper, in your own words
  3. Notice where you get stuck That is what you have not understood yet
  4. Ask about that point next

Then back to step 1: Ask for a simple explanation

Kind of factWhy the AI may be wrongWhere to check
Exam dates, fees, eligibility, vacanciesThey change every cycle and may be after its training cut-offThe official notification on the recruiting body's website
Rules, laws, constitutional provisionsNumbers and wording are easily mixed upThe official text or a standard textbook
Current affairsRecent events may be missing or confusedOfficial press releases and reliable newspapers
History, geography, science factsIt may blend similar factsNCERT and standard textbooks

Use AI to understand and practise; use the official source or textbook to confirm every fact.

Key takeaways: give the chatbot your days, hours and weak areas for a usable plan; ask for simpler explanations and then teach the idea back; treat generated questions as practice and verify their keys; check translated technical terms; and confirm every date, fee, rule and current-affairs fact from the official source or textbook.

Read the full lesson

Lesson 4

Limits of AI chatbots: hallucination, training cut-off, bias and keeping data safe

LimitWhat it meansExample
HallucinationThe model produces false information that sounds plausibleA made-up book title, court case or quotation
Training cut-offIts knowledge stops when its training data was collectedIt does not know about a recent scheme or result
Confident wrong answersThe tone sounds equally sure whether right or wrongAn arithmetic slip stated firmly
BiasIt repeats unfair patterns present in its training dataShowing only men when asked for a picture of an engineer

Cross-checking a chatbot's answer

  1. Pick out the checkable claims Names, numbers, dates, article or section numbers, quotations
  2. Government matters: official website Exams, schemes and rules - read the official website or notification
  3. Subject facts: NCERT and textbooks
  4. Confirm in two independent sources Two chatbots agreeing does not count
  5. Open any cited source yourself Confirm it exists and says what was claimed
  6. Recompute any arithmetic

Fluent is not the same as correct. Check facts in the official source or textbook, confirm in two reliable sources, and never share passwords, OTPs or ID numbers with a chatbot.

Safe to typeNot safe to type
Your own notes on the Mughal empire, asking for a summaryYour email password
A maths question from a practice bookAn OTP received on your phone
A paragraph to translate, with names removedYour Aadhaar or bank account number

Key takeaways: chatbots hallucinate, have a training cut-off, sound confident when wrong and can reflect bias. Cross-check claims in official sources and textbooks, use two independent sources, open any cited source and redo arithmetic. Never paste passwords, OTPs, Aadhaar, bank details or ID documents into a chatbot.

Read the full lesson

Lesson 5

Deepfakes, voice clones, plagiarism and fair use of AI

Where to lookPossible warning sign
Hands and fingersExtra, missing or merged fingers
Text in the pictureSignboards or labels with garbled, unreadable letters
Face and lipsLip movements out of step with the words; unnatural blinking
Edges and detailsBlurred or melting edges around hair, ears, earrings, glasses
Light and shadowShadows pointing different ways; skin too smooth and shiny
SourceNo original source, posted only by unknown accounts

Answering a suspicious call in a familiar voice

  1. Spot the red flags Urgency, money, secrecy, a story you cannot check on the call
  2. Hang up Do not pay or share anything during the call
  3. Call back on the saved number
  4. Tell a family member

Checking a viral "exam postponed" video

  1. Do not forward it
  2. Open the exam body's official website and notices
  3. Look for an official press release
  4. Nothing official? Treat it as false Keep preparing on the official schedule
Fair useNot fair
Asking AI to quiz you on a chapterSubmitting an AI-written essay as entirely your own
Getting feedback on your own draft, then rewriting it yourselfChanging a few words of AI text and calling it original
Stating that AI suggested an outline, where rules allow itHiding AI help where the rules require you to declare it

Judge the request, not the voice: urgency + money + secrecy = stop and call back. Confirm viral clips from official sources. Present only your own work as your own.

Key takeaways: deepfakes copy real people; look for visual flaws but rely on the official source. Answer voice-clone calls by hanging up and calling back on a saved number, and never share an OTP. Declare AI help honestly, and keep all devices out of the exam hall.

Read the full lesson

Privacy · Terms · Refunds · Delivery