Nearby lessons
47 of 49Prompt Engineering - Questions and Answers
Chapter 1 — Quick Questions
Q: What is a prompt in AI?
A: An instruction given to AI to generate output — A prompt is the instruction you give AI to get output.
Q: Why does AI need clear instructions?
A: Because AI cannot guess your intention and follows what you type — AI is a pattern follower, not a mind reader.
Q: What does a good prompt give you?
A: A focused and useful answer — A clear, specific prompt gives direction.
Q: Which is the best example of a clear prompt?
A: Explain IT jobs for freshers in simple English — Topic + audience + language make it clear.
Q: How does AI work step by step?
A: Type → read → understand → generate — You type, AI reads, AI understands, AI generates.
Q: What happens when you give an unclear prompt?
A: AI guesses and gives a weak answer — An unclear prompt makes AI fill the gaps with guesses.
Q: Prompting is best described as:
A: A communication skill — The better you explain, the better AI works — that is communication.
Q: The auto driver example teaches us that:
A: A clear instruction is the key to a correct result — Same driver, same auto — only the clear instruction changes the result.
Chapter 1 — Common Mistakes
- Writing one-word or very short unclear prompts like "Tell about job" — AI does not know what you want.
- Assuming AI will understand automatically — AI is not a mind reader; it fills gaps with guesses.
- Not mentioning the purpose — without a clear goal, the output becomes generic.
- Using confusing sentences and expecting a clear answer.
- Not mentioning the audience — the same topic needs different language for a school student vs a college student.
Chapter 2 — Quick Questions
Q: What are the two main clarity rules?
A: Specific and complete — A clear prompt is specific (focus) and complete (full information).
Q: What happens when a prompt is incomplete?
A: AI guesses and gives a generic output — Missing details mean AI fills the gaps with guesses.
Q: Which part of the formula tells AI who to act like?
A: Role — Role defines who AI should act like — teacher, HR, interviewer.
Q: Which part controls rules and limits like "no jokes"?
A: Constraints — Constraints are the rules and limits that control AI behavior.
Q: "Explain Java basics for freshers in simple English" is:
A: A specific prompt — Topic + audience + style are all clear, so it is specific.
Q: Why is the output format part important?
A: It decides how the answer looks — Output format gives shape — table, bullets, email, word count.
Q: The tailor example teaches that:
A: A structured instruction gives the exact result you want — Like a tailor, AI needs every detail in order.
Q: What is the prompt formula?
A: Role + Task + Context + Constraints + Output format — These five parts structure any professional prompt.
Q: What does the escalation ladder teach?
A: One prompt improving level by level, adding one formula part at a time — Raw → +Role → +Task → +Context → +Constraints → +Format. Each level makes the output more predictable.
Q: When comparing a bad prompt with a good prompt, what is the best teaching approach?
A: Keep the same topic so only the prompt quality changes — Same topic, two prompts — the only variable is the prompt itself, so the lesson is clear.
Chapter 2 — Common Mistakes
- Writing prompts without a role — AI gives a flat, direction-less answer.
- Missing the context — "for a fresher" changes the whole answer, but it is often left out.
- No constraints — AI adds jokes, greetings, and extra text you never asked for.
- Not defining the output format — you get a paragraph when you wanted a table.
- Writing random unstructured prompts and then blaming AI for a messy output.
- Not building prompts level by level — jumping straight to the final version skips the learning and hides why each part matters.
Chapter 3 — Quick Questions
Q: What does tone control in AI output?
A: How AI speaks — the style and feeling — Tone decides style, not content.
Q: Which tone is best for interview practice?
A: Strict — Interviews need direct, short, no-jokes answers.
Q: How do you stop AI from writing long answers?
A: Mention the length clearly in the prompt — Tell AI the size — 2 lines, 100 words, detailed.
Q: What is step-by-step prompting?
A: Asking AI to explain in ordered steps — Step-by-step gives a structured explanation.
Q: Which prompt combines all controls?
A: Explain the for loop step by step, each step in 2 lines, in simple English — It includes steps, length, and tone together.
Q: "Explain Python in 2 lines" controls what?
A: Length — The 2-line instruction sets the answer length.
Q: Why is daily practice important for prompting?
A: Because real skill comes from doing, not reading — Practice turns understanding into skill.
Q: Same question + different tone = ?
A: Different style, same topic — Tone changes the style, not the meaning.
Chapter 3 — Common Mistakes
- Not mentioning the tone and expecting AI to guess it — you get a random style.
- Using the wrong tone for the situation, like jokes in a formal office email.
- Mixing multiple tones in one prompt, so the output becomes confused.
- Not mentioning the length — AI writes an essay when you wanted two lines.
- Reading long unnecessary answers instead of controlling the size in the prompt.
- Not practicing daily — copying prompts without understanding them.
Chapter 4 — Quick Questions
Q: What does format control change in AI output?
A: How the answer is presented — Format changes presentation, not content.
Q: Which format is best for comparing Python and Java?
A: Table — Tables make comparison clear.
Q: What is structured output?
A: Forcing AI to follow a fixed format every time — Structured output = same headings, same order, same structure.
Q: Why is the word "strictly" useful in a prompt?
A: It stops AI from changing the format — "Strictly" locks the output to your structure.
Q: The "Only Output This" control is essential for:
A: JSON, apps and automation — Tools need clean data without extra text.
Q: Which phrase stops AI from adding extra explanation?
A: Do not add extra text — "Do not add extra text" forces only-output.
Q: A reusable prompt is:
A: One fixed prompt used for many topics — One fixed prompt, many topics, consistent output.
Q: The restaurant example teaches that:
A: A clear presentation instruction gives a perfect result — Same food — only the instruction about presentation changed.
Chapter 4 — Common Mistakes
- Not specifying the format — AI gives a messy paragraph instead of clean bullets.
- Accepting messy output instead of re-prompting with a format.
- Allowing AI to change the headings — without the word "strictly" the structure drifts.
- Not using the "only output" instruction — AI adds explanation that breaks tool data.
- Mixing explanation with data in one output instead of keeping them separate.
- Building prompts that cannot be reused — changing format every time wastes effort.
Chapter 5 — Quick Questions
Q: How does an LLM understand a prompt?
A: By following written instructions step by step — LLMs follow only written instructions, in order.
Q: Which word restricts the output?
A: only — "only" restricts — "give only bullet points".
Q: Why does AI give random answers sometimes?
A: Because the prompt is ambiguous — Ambiguity — missing details make AI guess.
Q: Which control word fixes a number?
A: exactly — "exactly 3 bullet points" fixes the count.
Q: What are constraints in a prompt?
A: Rules that limit AI behavior — Constraints are rules controlling length, tone, format.
Q: Why are examples useful in a prompt?
A: They guide AI to follow a specific pattern — Examples give AI a visible reference to copy.
Q: In the meeting example, which instruction dominates?
A: The last one — The last, clearest instruction wins.
Q: What is the best way to reduce randomness in AI output?
A: Use constraints and examples — Constraints + examples = control + stability.
Chapter 5 — Common Mistakes
- Writing instructions in random order — the role and task get lost in the middle.
- Giving conflicting instructions, like "be formal" then "make it funny".
- Ignoring small control words — casual prompts produce unpredictable output.
- Using vague words like "explain this" that force AI to guess.
- Not mentioning the audience — the same topic needs different language for different readers.
- Giving AI too much freedom — expecting consistent output without any constraints or examples.
Chapter 6 — Quick Questions
Q: What is zero-shot prompting?
A: Asking AI directly without any example — Zero-shot = direct asking, no example.
Q: What is the limitation of zero-shot prompting?
A: Output may vary and lack structure — Without an example, AI guesses and output varies.
Q: One-shot prompting uses how many examples?
A: One — One example guides AI's style.
Q: How many examples are ideal for few-shot prompting?
A: 2–5 — 2–5 examples give clear pattern learning.
Q: Few-shot prompting is most useful for:
A: Classification and pattern tasks — Few-shot teaches patterns — great for classification.
Q: Why is few-shot + structure powerful?
A: It gives consistent, reusable output — Few-shot teaches, structure locks — production-ready.
Q: Which prompt would extract clean JSON?
A: Extract details in JSON format only, using an example, no extra text — Example + JSON-only + no extra text = clean extraction.
Q: The tailor example belongs to which technique?
A: One-shot — Show one shirt → tailor copies the style → one-shot.
Chapter 6 — Common Mistakes
- Expecting structured output from a zero-shot prompt — no example, no format lock.
- Using zero-shot for automation tasks where stable format matters.
- Giving too many examples in few-shot — 2–5 is ideal, more creates confusion.
- Giving unclear or low-quality examples — AI copies bad patterns.
- Mixing different patterns in the examples, so AI cannot find one consistent style.
- Allowing extra text — forgetting "JSON only" or "no explanation" in structured output.
Chapter 7 — Quick Questions
Q: What does Role define in a prompt?
A: The job AI does — Role is the job — teacher, HR, interviewer.
Q: Persona controls which of these?
A: Behaviour — Persona controls how AI behaves.
Q: Tone defines what?
A: The voice style of communication — Tone is the voice feeling — polite, serious, funny.
Q: What are the three parts of persona control?
A: Behaviour, depth, attitude — Persona = behaviour + depth + attitude.
Q: What makes AI output stable and reusable?
A: Persona + constraints + format — The stability layer = persona + constraints + format.
Q: A domain persona makes AI:
A: Think from a specific field — Domain persona gives the right perspective — HR, analyst, CEO.
Q: What should a safety persona do when asked for illegal help?
A: Politely refuse and suggest a safe alternative — Safety persona refuses harmful requests gracefully.
Q: Same problem, different domain personas = ?
A: Different answers from different perspectives — Each domain focuses differently — HR, analyst, mentor differ.
Q: What does the persona ladder add at each level?
A: Role, then persona, then tone — Raw → +Role → +Persona → +Tone. Each level adds one layer of identity — the job, the behaviour, then the voice.
Chapter 7 — Common Mistakes
- Confusing role with persona — a role alone does not define behaviour.
- Not defining behaviour clearly — AI behaves randomly without a persona.
- Mixing multiple personas in one prompt, so the output loses its identity.
- Missing constraints or format in a persona prompt — output becomes unstable.
- Ignoring safety rules for risky topics like exams, hiring, or health.
- Using the wrong perspective — asking an analyst for people reasons or an HR for data reasons.
- Not building the persona level by level — mixing role, persona, and tone in one jump makes the output unstable.
Chapter 8 — Quick Questions
Q: What is a reasoning prompt?
A: A prompt that asks AI to think step by step — Reasoning prompts show AI's thinking process.
Q: Why is step-by-step thinking more trustworthy?
A: It shows the working, so mistakes are visible — The teacher trusts the student who shows steps.
Q: Manual Chain of Thought means:
A: You write the thinking steps in the prompt — Manual CoT = you control the thinking process.
Q: Auto-CoT means:
A: AI generates the reasoning examples automatically — Auto-CoT = AI creates the step-by-step examples.
Q: When should you use efficiency in prompting?
A: For simple questions that need quick answers — Simple questions get quick answers; complex ones get reasoning.
Q: Which example teaches Manual CoT?
A: The travel directions example — Step-by-step directions = manual CoT.
Q: What is the danger of not verifying an answer?
A: The answer may be wrong and you trust it blindly — Always verify important outputs.
Q: Normal prompt vs reasoning prompt — which reduces mistakes?
A: Reasoning prompt — Step-by-step reasoning reduces mistakes.
Chapter 8 — Common Mistakes
- Asking for direct answers without reasoning — AI rushes and makes mistakes.
- Using reasoning prompts for simple questions — wasted time and tokens.
- Not breaking complex problems into steps before asking.
- Ignoring the result — forgetting to verify the final answer.
- Writing unclear steps that confuse AI instead of guiding it.
- Letting AI choose the output length — long answers you did not ask for.
Chapter 9 — Quick Questions
Q: What is reflection prompting?
A: AI reviews and improves its own answer — Reflection = generate → review → fix → final.
Q: Self-consistency means:
A: Solving the problem multiple times and comparing — Multiple attempts reduce errors.
Q: Tree of Thoughts (ToT) is best for:
A: Decisions with multiple options — ToT compares options before choosing.
Q: In ToT, what should you do before comparing options?
A: List pros and cons for each option — Compare using the same criteria.
Q: Chain of Hindsight (CoH) means:
A: Learning from past mistakes and improving — Mistakes become improvements.
Q: Step-Back reasoning means:
A: Understanding the concept before solving — Concept first, then solution.
Q: What is the combined 'check twice, answer once' idea?
A: Reflection + self-consistency — Review and compare before trusting an answer.
Q: Which technique uses pros/cons and scores out of 10?
A: Tree of Thoughts — ToT compares options with scores and reasons.
Chapter 9 — Common Mistakes
- Trusting the first output blindly without asking AI to check it.
- Using only one solution when an important decision needs multiple tries.
- Considering only one option in a decision — ToT needs at least 3 branches.
- Comparing options with different criteria, so the comparison is unfair.
- Jumping straight into a solution without understanding the concept.
- Ignoring the mistake review step — no improvement happens.
Chapter 10 — Quick Questions
Q: Graph of Thoughts (GoT) means thinking in a:
A: Connected network — GoT = everything connected.
Q: What is the difference between ToT and GoT?
A: ToT is branches, GoT is a network — ToT branches out; GoT connects everything.
Q: The wedding-planning example teaches:
A: Everything is connected — Guests, budget, date — all connected.
Q: What is reliability engineering in AI?
A: Making AI safe, accurate, and consistent — Reliability = accuracy + safety + trust.
Q: Guardrails in a prompt are:
A: Allowed / not-allowed rules — Guardrails limit what AI may answer.
Q: What is a hallucination in AI?
A: AI creating fake facts with confidence — Hallucination = fake information.
Q: Which rule stops AI from creating fake facts?
A: Do not guess — "Do not guess" prevents invented facts.
Q: Complex case simulation combines:
A: CoT + ToT + Self-Consistency + Reflection — Real problems need multiple thinking methods.
Chapter 10 — Common Mistakes
- Thinking about only one factor — real problems are interconnected.
- Ignoring the connections between components in a decision.
- Building AI systems with no guardrails — harmful or random output.
- Allowing AI to guess and create fake facts (hallucination).
- Trusting AI output blindly without validation.
- Using only one thinking method for a complex real-world problem.
Chapter 11 — Quick Questions
Q: What does ReAct stand for?
A: Reason + Act — ReAct = Reason (think) + Act (do).
Q: In ReAct, what repeats until the goal is done?
A: The Reason → Act loop — Think → act → re-evaluate → repeat.
Q: When should AI use a tool?
A: When calculation, data, or validation is needed — Tools reduce mistakes where accuracy matters.
Q: The student example shows tools like:
A: Calculator, app, dictionary — Tools are helpers for correctness.
Q: What happens without a stop condition?
A: AI may loop forever, wasting time and money — Infinite loop = cost waste.
Q: What limits repeated failures?
A: A retry limit — Retry limit prevents endless repeats.
Q: A good stop rule for cleaning a room is:
A: Clean for 30 minutes, stop when clean — Clear time + clear condition = stop condition.
Q: Smart AI uses tools, not:
A: Guesses — Verify with tools instead of guessing.
Chapter 11 — Common Mistakes
- Asking only for answers and not defining any actions — AI never completes a task.
- Not defining the Reason/Act steps — the workflow has no process.
- Guessing numbers instead of using a calculator tool.
- Using tools unnecessarily for simple answers — wasted time.
- No stop condition — AI runs in an infinite, costly loop.
- Unlimited retries and no output limit — expensive and unstable.
Chapter 12 — Quick Questions
Q: What does a planning prompt do?
A: Breaks a big goal into a structured plan — Phases, tasks, priority, timeline, milestones.
Q: Which is a correct order for house construction?
A: Buy land, design, build, paint — Wrong order = failure; correct order = success.
Q: A milestone is:
A: A big checkpoint that measures progress — Milestones keep a project on time.
Q: An AI agent is best described as:
A: A system that thinks + acts + verifies + improves — It performs a full workflow.
Q: Which techniques does the mini agent combine?
A: CoT, ToT, CoH, Reflection, ReAct, guardrails, stop — All techniques work together in one agent.
Q: What does the capstone represent?
A: Combining all skills into a real-world system — Capstone = your final performance.
Q: The office assistant example is used to explain:
A: AI agents — The assistant plans, calculates, writes, and checks.
Q: Why include a stop condition in an agent?
A: So the agent finishes and returns the final output — Agents must stop after completion.
Chapter 12 — Common Mistakes
- Not breaking a big task into phases — the plan stays confusing.
- Ignoring the sequence — doing tasks in the wrong order.
- No priority — everything looks equally important.
- No timeline — the project never finishes on time.
- Building an agent without combining techniques — it becomes a plain prompt.
- Skipping validation, safety, or the stop condition in a workflow.
Chapter 13 — Quick Questions
Q: When AI gives a wrong output, who is usually at fault?
A: The prompt — AI is not failing — the prompt is failing.
Q: What is prompt refactoring?
A: Rewriting a messy prompt into clear structure — Same meaning, better clarity.
Q: Which is part of the prompt testing checklist?
A: Is the output format defined? — Role, task, format, ambiguity, data, testability.
Q: What is the improvement loop?
A: Ask → Check → Refine → Retest — Repeat until the output is correct.
Q: Why do we need multiple iterations?
A: To achieve accurate output — Each iteration improves the output.
Q: What is a meta-prompt?
A: A prompt that works on another prompt — Meta = prompt that analyzes and improves prompts.
Q: The debugging process is:
A: Diagnose → Explain → Rewrite → Test — Find, explain, rewrite, test.
Q: How should a prompt engineer handle a client's failing chatbot?
A: Analyze, identify, refactor, test, present — Apply the full debugging process.
Chapter 13 — Common Mistakes
- Jumping straight to the solution without diagnosing the problem.
- Writing one prompt and never refining it — the first try is never perfect.
- Not applying a testing checklist before delivering a prompt.
- Ignoring ambiguity or missing data in the prompt.
- Rewriting without structure — the fix repeats the same mistake.
- Skipping the final test after refactoring.
Chapter 14 — Quick Questions
Q: Why does a single prompt fail for big tasks?
A: It mixes tasks and loses structure — Big tasks need chaining, not one big prompt.
Q: What is prompt chaining?
A: Breaking one big task into small connected prompts — Output of one → input of next.
Q: The chai-shop example teaches that:
A: Order and steps decide the result — Wrong order = bad tea, right order = good tea.
Q: Decomposition uses which framework?
A: Input → Process → Output — Each step gets input, does one process, gives output.
Q: What is a linear pipeline?
A: A workflow where each step depends on the previous — Straight-line execution, no skipped order.
Q: Format locking means:
A: Keeping the same structure across steps — Consistency between steps keeps the pipeline stable.
Q: Which is the classic notes pipeline?
A: Chapter → Notes → Summary → Quiz — Each step feeds the next.
Q: What should each step in a pipeline produce?
A: A clear, useful output — Every step must have a defined output.
Chapter 14 — Common Mistakes
- Giving one big prompt with many tasks mixed together.
- Not defining the output of each step — outputs become unclear.
- Using the wrong order of steps — later steps fail on bad input.
- Adding new information in a later step instead of using only the previous output.
- Changing the format between steps — the pipeline loses consistency.
- Not testing each step before connecting them together.
Chapter 15 — Quick Questions
Q: What is a branching pipeline?
A: A workflow that changes path based on a condition — IF condition → Route A, ELSE → Route B.
Q: In the exam example, marks ≥ 80 lead to:
A: Advanced class — The condition decides the path.
Q: Topic drift means:
A: AI moving away from the main topic — Controlled chaining prevents drift.
Q: Which rule stops new information being added?
A: Do not add new information — Only the previous output is used.
Q: Variable injection uses:
A: Placeholders like {{TOPIC}} — Placeholders make prompts reusable.
Q: Why use variables in a prompt?
A: To make prompts dynamic and reusable — One template, many inputs.
Q: "Happy Birthday {{NAME}}" is an example of:
A: Variable injection — Change the variable, reuse the template.
Q: What is the benefit of branching pipelines?
A: Personalized output for different users — Different conditions → different paths → customization.
Chapter 15 — Common Mistakes
- Not defining the condition clearly — branching sends users down the wrong path.
- Mixing the outputs of both branches — the flows get confused.
- Letting AI drift off-topic in a multi-step workflow.
- Adding new information instead of using only the previous output.
- Hardcoding values instead of using variables — one use only.
- Using unclear variable names or no structure in a template.
Chapter 16 — Quick Questions
Q: What is a reusable workflow template?
A: A pre-built prompt system for many tasks — Build once, use many times.
Q: Workflow automation means:
A: Steps run automatically and pass data by themselves — From prompts to systems.
Q: Why is JSON used in workflows?
A: For structured data passing between steps — Tools and APIs read JSON automatically.
Q: What does the capstone project show?
A: Your real skill and learning — Capstone = from learner to system designer.
Q: Which is a capstone project option?
A: Resume Assistant or Study Assistant — Real problems with full workflows.
Q: Optimization in AI workflows means:
A: Improving speed, accuracy, cost, and stability — Production-ready systems.
Q: What should you do with unnecessary steps?
A: Remove them — Removing extra steps reduces cost.
Q: The school-template example teaches:
A: Same format → reusable — A fixed structure can be reused.
Chapter 16 — Common Mistakes
- Not modularizing prompts — the template cannot be reused.
- Hardcoding inputs instead of using variables.
- Not structuring output — JSON is skipped and data passing breaks.
- Not thinking about automation — manual repetitive work remains.
- Mixing steps or having no clear final output in the capstone.
- Keeping unnecessary steps — high token cost and slow workflows.
Chapter 17 — Quick Questions
Q: Prompt serialization means:
A: Structuring prompts using roles — Structured roles give controlled output.
Q: What does the system message control?
A: Behaviour — System = how AI behaves.
Q: Which message has higher priority?
A: System message — System instructions dominate.
Q: The Alpaca template uses which structure?
A: Instruction / Input / Output — Three parts: what to do, context, expected result.
Q: ChatML is best for:
A: Multi-turn conversations with context — ChatML brings conversation and memory.
Q: INST format is used by:
A: LLaMA and open-source models — Instruction in [INST] tags.
Q: Which format is best for building a chatbot?
A: ChatML — ChatML supports roles and context.
Q: Which format is best for dataset creation and training?
A: Alpaca — Alpaca's Instruction/Input/Output fits training.
Chapter 17 — Common Mistakes
- Using only plain prompts — no behaviour control.
- Mixing behaviour and task in one message instead of separating system and user.
- Not defining the system role — AI behaves randomly.
- Missing the Input part in Alpaca format — the task has no context.
- No context tracking in a multi-turn ChatML conversation.
- Using the wrong format for the model — e.g. INST tags on a model that expects ChatML.
Chapter 18 — Quick Questions
Q: What is prompt minimalism?
A: Writing short, clear prompts without losing meaning — Short + clear + complete.
Q: Prompt noise means:
A: Extra words that do not add value — Cut the fillers.
Q: What is a token?
A: A small piece of text AI processes — Tokens = building blocks of text.
Q: More tokens in a prompt means:
A: More cost and slower processing — Long prompts are expensive.
Q: A hallucination in AI is:
A: An incorrect but confident answer — AI creates fake facts with confidence.
Q: How do you reduce hallucination?
A: Ask assumptions, sources, and verification — Verify before trusting.
Q: Which rule allows AI to admit uncertainty?
A: Say "Not sure" when uncertain — "Not sure" stops fake confidence.
Q: What is the benefit of short prompts?
A: Low tokens, low cost, clear meaning — Less tokens = more efficiency.
Chapter 18 — Common Mistakes
- Writing long prompts with repeated meaning and polite fillers.
- Adding unnecessary words that create noise and confusion.
- Ignoring token count — expensive, slow prompts in real projects.
- Trusting AI output blindly without asking for verification.
- Not controlling the output format — inconsistent results.
- Allowing AI to answer confidently when it is not sure.
Chapter 19 — Quick Questions
Q: What is multimodal AI?
A: AI that works with multiple data types — Text + image + video + audio.
Q: What is the image prompt formula?
A: Subject + Environment + Lighting + Camera + Style — Each element adds quality to the image.
Q: Which element controls the atmosphere of an image?
A: Lighting — Lighting controls mood and atmosphere.
Q: What extra element does a video prompt need compared to an image?
A: Action + camera movement — Video = moving scene with motion.
Q: "A boy studying" is an example of:
A: A simple, weak image prompt — Short prompts give basic images.
Q: Which prompt would give a dramatic feel?
A: A superhero on a rooftop at night, dramatic lighting — Lighting + mood create drama.
Q: What is a transformation video?
A: One scene changing into another — Transformation = powerful for reels.
Q: The future belongs to creators who:
A: Combine creativity with AI — Multimodal tools reward creative prompting.
Chapter 19 — Common Mistakes
- Writing very short image prompts — *"a boy studying"* gives a flat, generic image.
- Missing the lighting — no atmosphere, no mood in the scene.
- No camera angle or style — the image has no perspective.
- Writing a video prompt like an image prompt — no action or camera movement.
- Ignoring details — the output stays basic and unprofessional.
- Not thinking like a director — the scene lacks emotion and motion.
Chapter 20 — Quick Questions
Q: What is negative prompting?
A: Telling AI what NOT to generate — What you don't want in the output.
Q: "Tea without sugar" is an analogy for:
A: Negative prompting — Tea = prompt, without sugar = negative prompt.
Q: Which word belongs in a negative prompt list?
A: Watermark — Watermark is an unwanted element.
Q: PAL stands for:
A: Program-Aided Language prompting — Write code and solve — Language → Code → Answer.
Q: What does PAL ask AI to do?
A: Write a program to solve the problem — Code proves the answer.
Q: PAL is NOT needed for:
A: Stories and marketing content — No logic is required there.
Q: Why is PAL more accurate than normal prompting?
A: Code runs real logic instead of guessing — Full logic, no guessing.
Q: Which is the basic PAL formula?
A: Write a Python program to + [problem] — Code + problem = accurate result.
Chapter 20 — Common Mistakes
- Asking AI directly for answers in maths and finance — it guesses.
- Not using negative prompts in image/video generation — blurry, distorted output.
- Forgetting common negative words like watermark, text, or extra fingers.
- Using PAL for stories and marketing — no logic is needed there.
- Not reusing the generated code — it is a reusable solution system.
- Trusting a guessed answer instead of asking AI to write code and prove it.
- Each Q&A comes straight from the chapter material
- Common mistakes are the exact errors beginners make
- Practise the MCQs, then try the full quiz page