Section II (Short Answer)
Q.2-i. How Does the Strong AI Differ from the Weak AI?
Answer:
- Weak AI (Narrow AI): Designed and trained for a specific, narrow task. It operates within pre-defined constraints and cannot think outside its programming. Example: Apple's Siri, chess-playing bots, or email spam filters.
- Strong AI (General AI / AGI): A theoretical form of AI that possesses human-level intelligence, consciousness, and the ability to understand, learn, and apply knowledge across any domain, exactly like a human being. Strong AI does not currently exist.
Q.2-ii. What are the misconceptions about Artificial Intelligence?
Answer:
Common misconceptions about AI include:
- AI has feelings and consciousness: People often confuse AI's ability to generate human-like text with actual understanding or emotion. AI operates purely on math and patterns.
- AI will replace all human jobs: While AI will automate repetitive tasks, it is more likely to transform jobs and create new tech-centric roles rather than cause total human unemployment.
- AI is completely objective and unbiased: AI learns from human-generated data; therefore, it can easily inherit and amplify human biases if the training data is flawed.
Q.2-iii. Why do we need Artificial Intelligence?
Answer:
We need AI primarily to handle tasks that exceed human capacity in terms of speed, scale, and pattern recognition.
- Automation: It frees humans from boring, repetitive tasks.
- Data Processing: AI can analyze massive datasets (Big Data) in seconds to extract valuable insights.
- Complex Problem Solving: It helps in critical fields like medical disease prediction, weather forecasting, and optimizing global supply chains.
Q.2-iv. How are Artificial Intelligence and Machine Learning related?
Answer:
Think of them as concentric circles. Artificial Intelligence (AI) is the broader, overarching concept of creating machines capable of simulating human intelligence. Machine Learning (ML) is a specific subset of AI. Instead of explicitly programming rules into a computer, ML provides the computer with data and algorithms, allowing it to "learn" the rules and improve its performance over time.
Q.2-v. Which is better for image classification? Supervised or unsupervised classification? Justify.
Answer:
Supervised classification is highly superior for image classification.
- Justification: Image classification requires the system to assign a specific label to an image (e.g., identifying whether an X-ray shows "pneumonia" or "healthy"). Supervised learning uses a training dataset where images are already clearly labeled by humans. The model learns the specific features of each label. Unsupervised learning, on the other hand, only groups images by similarity without knowing what they actually are, which doesn't fulfill the goal of explicit classification.
Q.2-vi. Which algorithm does Facebook use for face verification and how does it work?
Answer:
Facebook (Meta) primarily uses a deep learning system called DeepFace, which is based on a Convolutional Neural Network (CNN).
- How it works: When a photo is uploaded, the algorithm first detects faces and digitally aligns them to a standardized front-facing pose. It then passes the image through the CNN, which extracts specific facial features (like the distance between eyes or nose shape) and converts them into a mathematical 3D representation (a vector). It compares this vector to the vectors of your friends to suggest tags.
Q.2-vii. How can AI be used in detecting fraud?
Answer:
AI detects fraud by mastering pattern recognition. Banks feed AI models millions of historical transaction records. The AI learns the "normal" purchasing behavior of a user (e.g., typical locations, amounts, and times). When a new transaction occurs that drastically breaks this pattern—such as a sudden $5,000 purchase in a foreign country at 3 AM—the AI instantly flags it as an anomaly and blocks the transaction before it completes.
Q.2-viii. What is a Depth-first Search Algorithm?
Answer:
Depth-First Search (DFS) is an uninformed search algorithm used to traverse graphs or trees.
- Definition: It starts at the root node and explores as far deep as possible along each branch before backtracking.
- Mechanism: It uses a Stack data structure (Last-In-First-Out). It is highly memory efficient but is not guaranteed to find the shortest path.
- Example: Navigating a maze by keeping your hand on the left wall until you hit a dead end, then walking back to the last intersection to try the next path.
Section III (Essay Type)
Q.#.3. Give some real-world applications of AI and explain each?
Answer:
AI is heavily integrated into modern industries. Here are major real-world applications:
- Healthcare and Medicine:
- Explanation: AI is used for early disease prediction and medical imaging. AI models can scan MRIs and X-rays to detect early signs of tumors or pneumonia with higher accuracy than human radiologists.
- E-Commerce and Retail:
- Explanation: Platforms like Amazon use AI for recommendation engines. By analyzing your past purchases, search history, and time spent on pages, AI predicts what you are most likely to buy next and shows you personalized product feeds.
- Transportation and Autonomous Vehicles:
- Explanation: Companies like Tesla use computer vision and reinforcement learning to build self-driving cars. AI processes real-time data from cameras and radars to steer, brake, and navigate traffic safely without human intervention.
- Natural Language Processing (NLP) / Virtual Assistants:
- Explanation: Tools like ChatGPT, Alexa, or Google Assistant use AI to understand spoken or typed human language, process the context, and generate intelligent, conversational responses or perform tasks like setting alarms.
Q.#.4. What are some search strategies used in AI? Explain Breadth First Search in detail?
Answer:
Search Strategies in AI:
Search strategies are divided into two main categories:
- Uninformed Search (Blind Search): Searches without any domain knowledge (e.g., Breadth-First Search, Depth-First Search, Uniform Cost Search).
- Informed Search (Heuristic Search): Uses heuristics or estimates to find the goal faster (e.g., A* Search, Greedy Best-First Search).
Detailed Explanation of Breadth-First Search (BFS):
Breadth-First Search is an uninformed search algorithm that explores a tree or graph level by level.
-
How it Works: It starts at the root node (level 0). It then explores all the immediate neighbor nodes at the current depth level before moving on to the nodes at the next depth level.
-
Data Structure: It implements a Queue (First-In-First-Out / FIFO). The first node discovered is the first one to be expanded.
-
Properties:
-
Completeness: Yes, if a solution exists, BFS is guaranteed to find it.
-
Optimality: Yes, it is optimal for unweighted graphs (it always finds the shortest path in terms of the number of steps).
-
Example: If you are looking for a lost TV remote in your house, BFS means you first search the entire living room (level 1), then you search the entire kitchen (level 2), rather than searching the living room, then a bedroom, then the garage randomly.
Q.#.5. What are the intelligent agents in AI? Explain with examples and where are they used?
Answer:
Definition of an Intelligent Agent:
An intelligent agent is an autonomous entity in AI that observes its surrounding environment through sensors, makes decisions based on its programming or learned knowledge, and takes actions in that environment using actuators to achieve a specific goal.
Key Components:
- Sensors: How it gets data (cameras, microphones, thermometers).
- Actuators: How it affects the world (wheels, screens, robotic arms).
- Performance Measure: How its success is judged.
Examples and Where They Are Used:
- Self-Driving Car Agent:
- Where it's used: Automotive industry.
- Explanation: It uses cameras/LIDAR (sensors) to perceive the road, cars, and pedestrians. It calculates the safest route and uses the steering wheel and brakes (actuators) to reach the destination (goal).
- Smart Thermostat (e.g., Google Nest):
- Where it's used: Smart Home Automation.
- Explanation: It senses the room temperature (sensor). If the temperature drops below the user's preference, it triggers the heating system (actuator) to warm the house, learning the user's schedule over time.
- NPCs (Non-Player Characters) in Video Games:
- Where it's used: Gaming Industry.
- Explanation: An enemy guard in a game senses the player's presence visually or by sound (virtual sensors) and decides to attack or hide (virtual actuators) based on its health level.
Q.#.6. What are the different areas where AI has a great impact, explain with examples?
Answer:
AI is fundamentally changing how major sectors operate. The areas with the greatest impact include:
- Cybersecurity Sector:
- Impact: AI models monitor network traffic 24/7 to identify unusual data transfers or login attempts.
- Example: AI systems can detect a zero-day malware attack by recognizing its malicious behavior patterns before a human security analyst even realizes a breach is happening.
- Finance and Banking Sector:
- Impact: AI handles massive volumes of data for algorithmic trading, credit scoring, and customer service.
- Example: When you apply for a loan, AI evaluates your credit history, income, and spending habits in seconds to determine the risk of default, replacing the slow, manual human review process.
- Agriculture Sector:
- Impact: AI is driving "precision agriculture" to maximize crop yields while minimizing resource use.
- Example: AI-powered drones fly over fields taking pictures of crops. Computer vision algorithms analyze these images to identify early signs of plant disease or pest infestations, allowing farmers to spray pesticides only on the affected areas rather than the whole field.
- Education Sector:
- Impact: AI provides personalized learning experiences tailored to an individual student's pace.
- Example: Intelligent tutoring systems track a student's performance on math quizzes. If it notices the student struggles with fractions, it automatically adapts the curriculum to provide more practice and simpler explanations for that specific topic.