CBSE Class 12 Artificial Intelligence Previous Year Questions (843) with Solutions — 2025-26

Are you preparing for the CBSE Class 12 Artificial Intelligence (Subject Code 843) board exam? Solving Previous Year Questions (PYQs) is the most effective way to understand the exam pattern, know which topics are asked repeatedly, and practise writing board-quality answers. In this post, we have compiled important PYQs from CBSE Board Exams 2023, 2024, and 2025 — with complete, marking-scheme-aligned answers. All content is based on the official CBSE curriculum for Class 12 AI (Code 843), Session 2025-26.

AI Logic School — Free CBSE Resources

📋 Class 12 AI (Code 843) — Previous Year Questions

Important PYQs with Answers | Board Exam 2023, 2024 & 2025 | Session 2025-26

📘 Class 12 🤖 Subject Code 843 ⭐ PYQs with Answers 🎯 50 Marks Theory ⏱️ 2 Hours 🆓 Free

Class	12
Subject Code	843 — Artificial Intelligence
Content Type	Previous Year Questions (PYQs) with Answers
Years Covered	2023 | 2024 | 2025 Board Exams
Theory Marks	50 Marks | 2 Hours
Practical Marks	50 Marks (Project + Viva)
Source	CBSE Board Exams | academic.nic.in

🎯 Why Solve PYQs?

• Understand the exact exam pattern and question types
• Know which topics are repeatedly asked every year
• Practice time management for 2-hour paper
• Learn how to write answers as per CBSE marking scheme
• Build confidence before board exams

📊 Exam Pattern — Class 12 AI (843)

50

Theory Marks

Written Exam

50

Practical Marks

Project + Viva

21

Total Questions

Attempt 15

2 hrs

Duration

Theory Paper

Section	Type	Questions	Attempt	Marks
Section A	Objective Type (MCQ/Assertion-Reason)	5	All 5	24
Section B	Subjective Type (Short/Long Answer)	16	Any 10	26
Total				50

📚 Syllabus Units — Class 12 AI (843) 2025-26

Part A

Employability Skills

Communication, Self-Management, ICT Skills, Entrepreneurial Skills, Green Skills

Unit 1

Python Programming – II

Advanced Python, Functions, File Handling, Libraries

Unit 2

Data Science Methodology

Data Collection, Analysis, Visualization, Capstone Project

Unit 3

Making Machines See

Computer Vision, Image Processing, CNN, Object Detection

Unit 4

AI with Orange Tool

Data Mining, Classification, Clustering, Visual Analytics

Unit 5

Introduction to Big Data

Big Data concepts, Hadoop, Analytics, Applications

Unit 6

Generative AI

GANs, VAEs, Neural Networks, Creative AI Applications

Unit 7

AI Project Cycle

Problem Scoping, Data, Modelling, Evaluation, Deployment

Unit 8

Ethics in AI

Bias, Privacy, Fairness, Responsible AI, AI in Society

🔘 Section A — Objective Type PYQs (MCQ & Assertion-Reason)

⚠️ Exam Tip: Section A has 5 questions — attempt ALL. No negative marking. MCQs carry 1 mark each. Assertion-Reason questions also carry 1 mark.

📅 Board Exam 2025

1

Which type of neural network is effective for processing visual data and uses a three-dimensional arrangement to extract features from images?

(a) Recurrent Neural Network (RNN)

(b) Convolutional Neural Network (CNN) ✓

(c) Generative Adversarial Network

(d) Simple Neural Network

1 Mark

Answer: (b) Convolutional Neural Network (CNN)
CNN uses convolutional layers to extract spatial features from images. Its 3D arrangement (height × width × depth) makes it ideal for image classification, object detection, and computer vision tasks.

2

Variational Autoencoders (VAEs) are designed to learn from data. What are their two main components?

(a) Generator and Discriminator

(b) Encoder and Decoder ✓

(c) Input and Output Layer

(d) Training and Testing Module

1 Mark

Answer: (b) Encoder and Decoder
VAEs have two main parts: the Encoder compresses input data into a latent space representation, and the Decoder reconstructs the data from that representation. Together they learn to generate new data similar to training data.

3

Assertion-Reason: Assertion (A): Social media posts and images are examples of structured data.   Reason (R): Unstructured data does not follow a predefined format.

(a) Both A and R are true, R is correct explanation of A
(b) Both A and R are true, but R is NOT the correct explanation of A
(c) A is false, R is true
(d) Both A and R are false

1 Mark

Answer: (c) A is false, R is true
Social media posts and images are unstructured data — not structured data. So Assertion A is FALSE. However, Reason R is TRUE — unstructured data indeed does not follow a predefined format like rows/columns.

📅 Board Exam 2024

4

Which step of the AI Project Cycle involves analysing data to discover hidden patterns and useful information?

(a) Problem Scoping

(b) Data Acquisition

(c) Data Exploration ✓

(d) Model Evaluation

1 Mark

Answer: (c) Data Exploration
Data Exploration is the 3rd step of the AI Project Cycle. It involves analysing collected data to discover hidden patterns, interesting insights, and useful information. Data is visualised in user-friendly formats like charts and graphs.

5

When a machine is able to mimic human traits and intelligence, it is said to be — (Assertion-Reason type)

1 Mark

Answer: Artificially Intelligent
When a machine mimics human cognitive traits such as learning, reasoning, problem-solving, perception and language understanding, it is called Artificial Intelligence. This is the fundamental definition of AI.

📝 Section B — Subjective Type PYQs

⚠️ Exam Tip: Section B has 16 questions — attempt ANY 10. Choose questions you are most confident about. Mix short and long answers for best marks.

📌 1 Mark Questions (Answer in 1 line)

📅 Board Exam 2025 & 2024

1

Define the term 'Data Features' in the context of AI and Data Science.

1 Mark

Answer: Data features are individual measurable properties or characteristics of a dataset used to analyse and build AI models. They provide context and information about the data (also called variables or attributes).

2

What is meant by 'Sustainable Agriculture' in the context of AI applications?

1 Mark

Answer: Sustainable agriculture is an environment-friendly farming approach that prevents excessive use of chemical fertilizers to protect soil health. AI helps make farming sustainable by predicting crop yield, water usage, and soil conditions.

3

Name the three fundamental layers of an Artificial Neural Network (ANN).

1 Mark

Answer: The three fundamental layers of an ANN are:
1. Input Layer — receives raw data
2. Hidden Layer(s) — processes and learns patterns
3. Output Layer — produces the final result/prediction

4

What is the role of a Discriminator in a Generative Adversarial Network (GAN)?

1 Mark

Answer: The Discriminator in a GAN acts as a classifier that distinguishes between real data and fake data generated by the Generator. It provides feedback to the Generator to improve its output quality.

📌 2 Mark Questions (Answer in 20-30 words)

1

Explain the difference between Supervised Learning and Unsupervised Learning with one example each.

2 Marks

Supervised Learning: The model is trained on labelled data (input-output pairs). The algorithm learns to map inputs to correct outputs.
Example: Email spam detection — emails are labelled as spam/not spam.

Unsupervised Learning: The model is trained on unlabelled data and finds hidden patterns on its own.
Example: Customer segmentation — grouping customers by buying behaviour without predefined labels.

2

What is NLP (Natural Language Processing)? Give two real-life applications of NLP.

2 Marks

NLP (Natural Language Processing) is a branch of AI that enables computers to understand, interpret, and generate human language (text and speech).

Real-life Applications:
1. Virtual Assistants — Siri, Alexa, Google Assistant understand spoken commands
2. Machine Translation — Google Translate converts text between languages automatically

3

Differentiate between Structured Data and Unstructured Data. Give one example of each.

2 Marks

Structured Data	Unstructured Data
Follows a predefined format (rows/columns)	No predefined format or organisation
Easy to search and analyse	Difficult to analyse directly
Example: Excel spreadsheet, SQL database	Example: Social media posts, images, videos

4

What is Computer Vision? Mention any two applications used in India.

2 Marks

Computer Vision is a field of AI that enables computers to interpret and understand visual information from images and videos, similar to how humans see.

Applications in India:
1. Aadhaar Face Authentication — facial recognition for identity verification
2. Traffic surveillance cameras — detecting vehicles, number plates, and traffic violations automatically

📌 3 Mark Questions (Answer in 40-60 words)

1

Explain the AI Project Cycle with its five steps. How does it help in building a real-world AI solution?

3 Marks

The AI Project Cycle is a systematic approach to building AI solutions:

Step 1 — Problem Scoping: Define the problem clearly — what AI needs to solve, who are the users, what data is needed.
Step 2 — Data Acquisition: Collect relevant, accurate data from various sources (sensors, surveys, internet).
Step 3 — Data Exploration: Analyse data to find patterns, remove errors, visualise trends.
Step 4 — Modelling: Select and train an appropriate AI/ML model on the data.
Step 5 — Evaluation: Test the model's accuracy and deploy it after validation.

This cycle ensures a structured, repeatable approach to building reliable AI solutions.

2

What is Big Data? Explain any three characteristics of Big Data (3 Vs).

3 Marks

Big Data refers to extremely large datasets that cannot be processed by traditional data management tools due to their size, speed, and variety.

Three Vs of Big Data:
1. Volume — Huge amount of data generated every second (e.g., millions of tweets, transactions)
2. Velocity — Data is generated at very high speed and needs real-time processing (e.g., stock market data)
3. Variety — Data comes in many formats: text, images, video, audio, structured and unstructured

3

A company is developing a smart security camera that identifies people and vehicles. Which AI domain is being used? Explain how it works.

3 Marks

AI Domain Used: Computer Vision

How it works:
1. The camera captures frames (images) continuously from the environment
2. A Convolutional Neural Network (CNN) analyses each frame
3. The model identifies and classifies objects — people, vehicles, etc. — by matching patterns learned during training
4. When an unknown person or suspicious vehicle is detected, an alert is triggered automatically

This uses Object Detection — a computer vision technique that locates and identifies multiple objects in an image in real time.

📌 4-5 Mark Questions (Detailed Answers)

1

Describe the structure of an Artificial Neural Network (ANN). Explain how it learns from data using the concept of weights and activation functions. (5 marks)

5 Marks

Structure of ANN:
An ANN is inspired by the human brain and consists of interconnected nodes (neurons) arranged in layers:

1. Input Layer: Receives raw data. Each neuron represents one feature/input variable.
2. Hidden Layer(s): One or more layers that process inputs. Each neuron applies a mathematical transformation.
3. Output Layer: Produces the final prediction or classification result.

How ANN Learns:
• Each connection between neurons has a weight — a numerical value that determines the strength of the connection
• Each neuron applies an Activation Function (e.g., ReLU, Sigmoid) to decide whether to pass the signal forward
• During training, the network compares its output to the correct answer and calculates error (loss)
• Using Backpropagation, the error is sent backwards through the network
• Weights are adjusted using Gradient Descent to minimise the error
• This process repeats over many iterations (epochs) until the network learns to make accurate predictions

2

What are ethical concerns in AI? Explain any four with examples from real life. (4 marks)

4 Marks

Ethical Concerns in AI:

1. Bias and Discrimination
AI systems can inherit biases from training data. Example: A hiring AI trained on historical data may unfairly reject women or minorities.

2. Privacy Violation
AI systems collect and analyse massive personal data without user consent. Example: Facial recognition cameras tracking citizens without permission.

3. Lack of Transparency (Black Box)
Many AI models cannot explain their decisions. Example: A bank AI rejects a loan application without giving any reason.

4. Job Displacement
AI automation replaces human jobs in manufacturing, banking, and customer service. Example: Chatbots replacing call centre employees.

4. Deepfakes and Misinformation
Generative AI creates fake images, videos, and voices. Example: Deepfake videos of politicians saying things they never said, spreading misinformation.

⭐ Most Important Topics — Asked Every Year

Topic	Unit	Frequency
AI Project Cycle (5 Steps)	Unit 7	⭐⭐⭐ Very High
Neural Networks (ANN, CNN)	Unit 3	⭐⭐⭐ Very High
Structured vs Unstructured Data	Unit 2	⭐⭐⭐ Very High
Supervised vs Unsupervised Learning	Unit 2	⭐⭐⭐ Very High
Generative AI (GANs, VAEs)	Unit 6	⭐⭐ High
Ethics in AI	Unit 8	⭐⭐ High
Computer Vision & NLP	Unit 3	⭐⭐ High
Big Data & its Characteristics	Unit 5	⭐⭐ High
Python Programming (functions, loops)	Unit 1	⭐⭐ High
Data Visualisation Tools	Unit 2, 4	⭐ Moderate

🔁 Quick Revision — Key Definitions

📌 Must-Know Definitions for Board Exam

• Artificial Intelligence: Technology that enables machines to mimic human intelligence — learning, reasoning, and problem-solving
• Machine Learning: A subset of AI where systems learn from data and improve without being explicitly programmed
• Deep Learning: A subset of ML using multi-layered neural networks to learn complex patterns
• CNN: Convolutional Neural Network — specialised for image and visual data processing
• GAN: Generative Adversarial Network — Generator + Discriminator working against each other to create realistic synthetic data
• VAE: Variational Autoencoder — Encoder + Decoder that learns to generate new data
• NLP: Natural Language Processing — AI that understands and generates human language
• Big Data (3 Vs): Volume, Velocity, Variety
• Data Features: Individual measurable properties of a dataset used to build AI models
• Overfitting: When a model learns training data too well and fails on new data

🇮🇳

💡 Did You Know?

India's UPI payment system uses AI and Machine Learning to detect fraudulent transactions in real time — processing over 10 billion transactions every month! This is a perfect real-world example of AI Project Cycle, Data Science, and Ethics in AI — all topics in your Class 12 board exam! 🚀

📋 Quick Answer Key — Section A MCQs

Q1 (2025)	(b) Convolutional Neural Network (CNN)
Q2 (2025)	(b) Encoder and Decoder
Q3 (2025 A-R)	(c) A is false, R is true
Q4 (2024)	(c) Data Exploration
Q5 (2024)	Artificially Intelligent

🔗 Related Posts on AI Logic School

🐍

Class 12 AI — Python Practical File

Programs with output + Viva Questions

📊

Data Science Methodology Notes

Complete chapter notes Class 12

🔢

Class 12 AI MCQ Bank

100+ MCQs chapter-wise with answers

🧠 AI Logic School

Free CBSE AI & Computational Thinking Resources | Classes 3–12

🌐 ailogicschool.blogspot.com  |  📢 t.me/ailogicschool  |  📸 @ailogicschool

© AI Logic School 2026 | Content based on CBSE official curriculum | For educational use only

Continue learning

📖

Read next chapter

Chapter 3: AI & Logic Gates

›

📥

Download worksheet

Free PDF practice sheet

›

🧠

Take the quiz

Test your AI basics

›

✈️

Join our Telegram channel

Get every new lesson in your Telegram

Join free