Artificial Intelligence (AI) MCQ Questions and Answers

Artificial Intelligence (AI) MCQ Questions and Answers

41. What is Deep Learning and how does it differ from traditional Machine Learning?

  1. Deep Learning uses more training data; traditional ML uses less data — there is no other difference
  2. Deep Learning uses multi-layered neural networks to automatically extract features from raw data; traditional ML requires manual feature engineering by domain experts
  3. Traditional ML is always more accurate than Deep Learning for all tasks
  4. Deep Learning only works on text data; traditional ML works on all data types

Answer : B
Explanation: The key difference is feature extraction. Traditional ML requires humans to manually identify and engineer relevant features from raw data before training. Deep Learning uses multi-layered neural networks that automatically learn hierarchical feature representations directly from raw data (images, text, audio). This eliminates manual feature engineering but requires much more data and computational power. Deep Learning excels at image recognition, speech recognition, and NLP, while traditional ML often works better with smaller, structured datasets.

42. What is a Convolutional Neural Network (CNN) used for in AI?

  1. A type of neural network used primarily for processing sequential time-series data
  2. A type of deep neural network designed specifically for processing grid-like data such as images, using convolutional layers to automatically detect spatial features
  3. A neural network that convolves multiple training datasets to produce better predictions
  4. A network architecture used for natural language processing and text generation tasks

Answer : B
Explanation: A Convolutional Neural Network (CNN) is a specialized deep learning architecture designed for processing visual data. It uses convolutional layers to automatically detect local features (edges, textures, shapes) in images without manual feature engineering. Key components: Convolutional Layer (applies filters to detect features), Pooling Layer (reduces spatial dimensions), Fully Connected Layer (makes final classification). CNNs power image classification, object detection (YOLO), face recognition, medical image analysis, and self-driving car vision systems. Famous CNN architectures include LeNet, AlexNet, VGG, ResNet, and InceptionNet.

43. What is a Recurrent Neural Network (RNN) and what is it used for?

  1. A neural network that recurrently retrains itself on the same dataset to improve accuracy
  2. A type of neural network designed for sequential data where connections form directed cycles, allowing information to persist across time steps
  3. A network that repeatedly applies the same convolutional filter across all image regions
  4. A neural network architecture used exclusively for image generation tasks

Answer : B
Explanation: A Recurrent Neural Network (RNN) is designed to work with sequential data where the order matters — like time series, speech, or text. Unlike standard neural networks, RNNs have connections that loop back, allowing them to maintain a “memory” of previous inputs. However, standard RNNs suffer from the vanishing gradient problem for long sequences. LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit) are improved variants that handle long-range dependencies. RNNs are used for language modeling, machine translation, speech recognition, and sentiment analysis.

44. What is the role of an activation function in a neural network?

  1. To activate the neural network’s training process when data is fed in
  2. To introduce non-linearity into the network’s output, enabling it to learn complex patterns that cannot be captured by linear transformations alone
  3. To normalize input data before it enters the neural network layers
  4. To determine the learning rate during the backpropagation training process

Answer : B
Explanation: Activation functions introduce non-linearity into neural networks, allowing them to learn complex, non-linear patterns in data. Without activation functions, a neural network would simply be a linear transformation regardless of depth. Common activation functions: ReLU (Rectified Linear Unit) — most widely used, outputs max(0, x). Sigmoid — outputs values between 0 and 1, used for binary classification output. Tanh — outputs between -1 and 1. Softmax — converts raw scores to probabilities for multi-class classification. Leaky ReLU — fixes the “dying ReLU” problem. Each activation function has specific use cases and properties.

45. What is backpropagation in neural network training?

  1. A process where training data is fed backwards through the network from output to input
  2. An algorithm that calculates the gradient of the loss function with respect to each weight and propagates errors backward through the network to update weights during training
  3. A technique for backing up the neural network’s weights during training checkpoints
  4. A process of removing unnecessary neurons from a neural network to reduce its size

Answer : B
Explanation: Backpropagation (backward propagation of errors) is the core algorithm for training neural networks. During the forward pass, input data flows through the network and a loss (error) is calculated. During backpropagation, the chain rule of calculus is used to compute the gradient of the loss with respect to each weight, propagating error signals backward from the output layer through all hidden layers. These gradients are then used by an optimizer (like gradient descent or Adam) to update the weights to minimize the loss. Backpropagation is what makes deep learning training possible.

46. What is a Large Language Model (LLM) in AI?

  1. A database system that stores large amounts of language data for AI retrieval
  2. A type of deep learning model trained on massive text datasets that can understand context and generate human-like text by predicting the most likely next token in a sequence
  3. A programming language interpreter designed for large-scale AI applications
  4. A dictionary system that maps words to their mathematical vector representations

Answer : B
Explanation: A Large Language Model (LLM) is a deep learning model — typically based on the Transformer architecture — trained on hundreds of billions of text tokens. LLMs learn statistical patterns of language and can generate coherent, contextually relevant text. Examples include GPT-4/GPT-5 (OpenAI), Claude (Anthropic), Gemini (Google), and Llama (Meta). LLMs power applications including chatbots (ChatGPT), code generation (GitHub Copilot), document summarization, question answering, and content creation. They represent the defining AI technology of the 2020s.

47. What is Generative AI?

  1. AI that generates reports about other AI systems’ performance
  2. A category of AI that can create new content — including text, images, audio, video, and code — that resembles human-created content, by learning patterns from training data
  3. AI systems that generate only numerical outputs from structured datasets
  4. A subset of AI focused specifically on generating training datasets for other AI models

Answer : B
Explanation: Generative AI refers to AI systems capable of generating new, original content by learning statistical patterns from training data. Types include: Text generation (ChatGPT, Claude), Image generation (DALL-E, Midjourney, Stable Diffusion), Music generation (Suno), Video generation (Sora), Code generation (GitHub Copilot). Key architectures include Transformers (for text), GANs (Generative Adversarial Networks), VAEs (Variational Autoencoders), and Diffusion Models. Generative AI has transformed creative industries, software development, education, and content creation since 2022.

48. What is the Transformer architecture in AI?

  1. A type of robotic transformer that converts energy for AI-powered machinery
  2. A deep learning architecture that uses self-attention mechanisms to process sequential data in parallel, overcoming the limitations of RNNs and enabling modern LLMs
  3. A traditional neural network architecture that transforms input images into text descriptions
  4. A data transformation pipeline used to preprocess raw data before AI model training

Answer : B
Explanation: The Transformer architecture, introduced in the 2017 paper “Attention Is All You Need” by Google researchers, revolutionized AI. Unlike RNNs that process sequences step by step, Transformers use self-attention mechanisms to consider all input tokens simultaneously — enabling parallel processing and capturing long-range dependencies far better. The core components are multi-head self-attention, positional encoding, and feed-forward layers. All modern LLMs (GPT, Claude, Gemini, Llama) are built on Transformer architecture. BERT (encoder-only), GPT (decoder-only), and T5 (encoder-decoder) are key Transformer variants.

49. What is a Knowledge Base in AI?

  1. A database that stores only numerical training data for machine learning models
  2. A structured repository of facts, rules, and relationships about a domain that an AI system uses to reason, make decisions, and answer queries
  3. A collection of pre-trained AI models stored for later deployment
  4. The main memory of a computer system used by AI algorithms during execution

Answer : B
Explanation: A Knowledge Base (KB) is a centralized repository of structured information that an AI system uses for reasoning and problem-solving. In expert systems, the KB consists of domain-specific facts and IF-THEN rules. In knowledge graphs (like Google’s Knowledge Graph), entities and their relationships are stored as triples (subject-predicate-object). Modern AI systems like RAG (Retrieval Augmented Generation) combine LLMs with knowledge bases to produce more accurate and verifiable answers. The KB works with an inference engine to derive conclusions from stored knowledge.

50. What is the difference between Narrow AI and General AI?

  1. Narrow AI is used in narrow (small) companies; General AI is used in large enterprises
  2. Narrow AI (Weak AI) is designed and trained for one specific task; General AI (AGI) would have human-level intelligence across any task — General AI does not currently exist
  3. General AI is less powerful than Narrow AI but more widely deployed
  4. Narrow AI requires more computing power than General AI for all applications

Answer : B
Explanation: Narrow AI (Weak AI) is AI designed and optimized for a specific task — ALL currently deployed AI systems are Narrow AI. Examples: ChatGPT (text generation), AlphaGo (playing Go), Siri (voice assistant), Tesla Autopilot (self-driving). Artificial General Intelligence (AGI) would possess human-level intelligence, able to learn and perform any intellectual task a human can — it does not yet exist and remains a research goal. Artificial Superintelligence (ASI) would surpass human intelligence in all domains — purely theoretical. This distinction is critical for AI interviews and exams.

Pages: 1 2 3 4 5 6 7 8 9 10