Generative AI MCQs - Class 12 AI (843)

Score higher with Generative AI MCQs ! Updated as per latest syllabus, with answers to help you revise faster and build strong concepts. Start now!

Q1. A company wants to develop an AI application that creates new content in different formats. Which combination of content can Generative AI produce?
a) Only text and images
b) Only audio and video
c) Text, images, audio and video
d) Only numerical data

Q2. Which of the following is NOT a Generative AI example?
a) Claude
b) DALL-E
c) Random Forest
d) Gemini

Q3. A VAE processes a dataset and converts the information into a compressed representation. What is this representation called?
a) Training boundary
b) Latent space
c) Classification layer
d) Output category

Q4. A company needs a generative model for producing highly realistic visual outputs. Based on the comparison in the chapter, which model would be a suitable choice?
a) GAN
b) VAE
c) Logistic Regression
d) Decision Tree

Q5. A research team needs a model that provides structured data generation and an interpretable latent space. Which model should it consider?
a) GAN
b) VAE
c) SVM
d) Random Forest

Q6. An AI system analyses the distribution of existing data and creates new samples that follow similar patterns. Which category of machine learning models does it represent?
a) Discriminative models
b) Generative models
c) Classification-only models
d) Rule-based models

Q7. The component of a VAE that translates information from the latent space back to its original form is the:
a) Encoder
b) Discriminator
c) Decoder
d) Generator

Q8. Discriminative models mainly focus on:
a) Defining class boundaries within the data
b) Creating new artworks
c) Generating new stories
d) Producing audio samples

Q9. A model learns that certain words and phrases separate spam emails from non-spam emails. This model is:
a) Generative
b) Discriminative
c) A VAE
d) A GAN

Q10. Which of the following is a discriminative algorithm?
a) Naïve Bayes
b) GAN
c) Support Vector Machine
d) Autoregressive model

Q11. Which group contains only generative algorithms or models?
a) Decision Trees, SVM
b) Logistic Regression, Random Forest
c) SVM, VAEs
d) Naïve Bayes, GAN, VAEs

Q12. The algorithms used to handle huge volumes of data are rooted mainly in:
a) Geometry and trigonometry
b) Calculus, probability and statistics
c) Algebra and mensuration
d) Set theory only

Q13. Which of the following is NOT an example of image generation?
a) Stable Diffusion
b) Stability AI
c) DALL-E
d) Meta AI’s Voicebox

Q14. Which is an example of a text generation tool?
a) Perplexity
b) Lumiere
c) MusicLM
d) Canva

Q15. Google’s Lumiere is an example of:
a) Audio generation
b) Text generation
c) Video generation
d) Image classification

Q16. Which pair is correctly matched?
a) Lumiere – audio generation
b) MusicLM – audio generation
c) Voicebox – video generation
d) Stable Diffusion – text generation

Q17. A Large Language Model (LLM) can perform which of the following NLP tasks?
a) Only image editing
b) Only data storage
c) Generating and classifying text, answering questions, translating languages
d) Only speech recording

Q18. LLMs are called “large” because they:
a) Occupy large physical space
b) Are trained on massive datasets of text and code
c) Can only be used by large companies
d) Have large screens

Q19. The performance of a language model is affected by:
a) The colour of its interface
b) The number of users
c) The quality of its dataset
d) The programming language of the website

Q20. Transformers are:
a) A type of neural network architecture that revolutionised NLP
b) A type of database
c) A discriminative algorithm for spam detection
d) A tool for video editing

Q21. An employee frequently writes emails and wants an AI application to suggest sentence continuations automatically. Which LLM application can help?
a) Email auto-completion
b) Video manipulation
c) Facial recognition
d) Anomaly detection

Q22. An AI system generates a script, which is then converted into natural-sounding spoken words by another system. Which technology is responsible for converting the text into speech?
a) Text-to-Speech (TTS)
b) Image-to-Text
c) Data Classification
d) Decision Tree

Q23. How do LLMs contribute to audio generation?
a) They directly generate audio signals
b) They record audio from the microphone
c) They generate text scripts that TTS systems convert into speech
d) They compress audio files

Q24. LLMs contribute to image-related tasks mainly through:
a) Image captioning
b) Directly drawing images
c) Photo editing
d) Image compression

Q25. For video content, LLMs can help by generating:
a) Video frames directly
b) Subtitles, captions or scene summaries
c) Camera movements
d) Video file formats

Q26. A school chatbot confidently gives a wrong date for an event. Which LLM limitation does this show?
a) High response time
b) Prioritising natural language over accuracy
c) Memorising specific details
d) Data privacy leakage

Q27. High response time and high cost in LLMs arise because:
a) Text is stored in the cloud
b) Processing text needs significant computational resources
c) LLMs use no datasets
d) LLMs cannot process text

Q28. The future of Generative AI places strong emphasis on:
a) Removing human collaboration
b) Ethical development to minimise biases and ensure responsible use
c) Reducing multilingual translation
d) Avoiding healthcare and education

Q29. DeepFaceLab and FaceSwap are examples of:
a) Text generation tools
b) Audio generation tools
c) Deepfake technology
d) Discriminative algorithms

Q30. Deepfake technology mainly threatens:
a) Internet speed
b) Trust in media integrity, by spreading misinformation
c) Battery life
d) Storage capacity

Q31. Superimposing a person’s face onto adult content without consent leads to:
a) Privacy violations and reputational damage
b) Better image quality
c) Improved data security
d) Reduced misinformation

Q32. Presenting AI-generated content as one’s own work mainly raises concerns about:
a) Internet speed
b) Hardware compatibility
c) Intellectual property rights and academic integrity
d) Colour accuracy

Q33. Why is disclosing the use of generative AI important?
a) It increases the speed of AI
b) It maintains trust and accountability
c) It reduces the cost of tools
d) It prevents the AI from learning

Q34. While citing AI-generated text, which should be placed in the author spot?
a) The student’s name
b) The teacher’s name
c) The tool name and “Generative AI tool”
d) The website’s owner

Q35. Which date should be used in the citation of AI-generated content?
a) The date the tool was released
b) The date you received the AI-generated content
c) The date of the school’s exam
d) The date the AI company was founded

Q36. For signing up to Canva for Education, verification is done through:
a) A phone call from Canva
b) A postal letter
c) An email sent to your school email address
d) A video interview

Q37. Which tool is used in the unit for AI video generation from a descriptive prompt?
a) Animaker
b) Canva Text to Image
c) Google AI Studio
d) MusicLM

Q38. The first step in creating a chatbot with the Gemini API is to:
a) Install a video editor
b) Obtain an API key from Google AI Studio
c) Create a deepfake
d) Write the exit condition

Q39. Which command installs the Python SDK for the Gemini API?
a) pip install -q -U google-generativeai
b) pip install canva
c) import chat
d) model.start_chat()

Q40. In the chatbot code, which statement correctly starts a chat session with an empty history?
a) chat = model.start_chat(history=[])
b) chat = genai.configure(history=[])
c) chat = input(history=[])
d) chat = model.exit(history=[])