470 Final Exam Flashcards
(59 cards)
Q: Generative AI can only create text, but not images or audio. (True/False)
False
Q: Which of the following is NOT a characteristic of human intelligence simulated by AI? a) Learning and Adaptation, b) Perception and Pattern Recognition, c) Superhuman Strength, d) Reasoning and Problem-Solving.
Superhuman Strength
Q: Superintelligent AI refers to an AI that surpasses human intelligence, including creativity and problem-solving abilities. (True/False)
True
Q: Of the 3 Types of AI, ___________ AI is capable of understanding, learning, and applying knowledge across a wide range of tasks, but it remains largely theoretical.
General AI or AGI
Q: Deep learning reduces the need for manual feature engineering because it can automatically learn features from large datasets. (True/False)
True
Q: Which of the following is a type of deep learning architecture/model? a) Decision Trees, b) Convolutional Neural Networks (CNNs), c) Support Vector Machines (SVMs), d) K-Nearest Neighbors (KNN).
Convolutional Neural Networks (CNNs)
Q: Which of the following is a common use case for traditional machine learning? a) Spam Detection, b) Image Generation, c) 3D Model Creation, d) Video Editing.
Spam Detection
Q: Arthur Samuel’s checkers program learned by playing against itself and improved over time without human intervention. (True/False)
True
Q: Discriminative models are used to generate new content, such as text or images. (True/False)
False
Q: Of the 3 types of AI, ___________ AI is where we currently are, focusing on AI that performs specific tasks.
Narrow or Weak AI
Q: Generative AI models are primarily used for classification tasks, like spam detection. (True/False)
False
Q: Which of the following tasks is NOT typically associated with Generative AI? a) Text Generation, b) Object/Image Classification, c) Video Generation, d) Music Creation.
Object/Image Classification
Q: Alan Turing introduced a famous test known as the ___________, which assesses a machine’s ability to imitate human conversation.
Turing Test
Q: The first AI Winter occurred in the 1990s due to unmet expectations and the limitations of expert systems. (True/False)
False
Q: Which of the following was one of the earliest AI systems that improved its strategy by playing against itself? a) Deep Blue, b) AlphaGo, c) Arthur Samuel’s Checkers Program, d) ELIZA.
Arthur Samuel’s Checkers Program
Q: The backpropagation algorithm is used to adjust weights in neural networks to minimize errors during training. (True/False)
True
Q: ___________ networks/models are designed to overcome the vanishing gradient problem by maintaining information over long sequences in tasks like language processing and are the successor to RNNs.
Long Short-Term Memory (LSTM)
Q: Alan Turing introduced the idea of genetic algorithms in his 1948 paper ‘Intelligent Machinery’. (True/False)
True
Q: Machine learning involves computers learning from data without being explicitly programmed. (True/False)
True
Q: What is the key innovation in transformer models, which has led to our recent explosion in LLMs such as ChatGPT? a) Recurrent processing, b) Self-attention mechanism, c) Gradient descent, d) Convolution layers.
Self attention mechanism
Q: The perceptron, developed by Frank Rosenblatt, used a simplified model of a biological neuron to make classification decisions. (True/False)
True
Q: The perceptron used a ___________ decision boundary to classify input data based on weights and inputs.
Linear
Q: Which neural network architecture is widely used for image recognition and classification? a) Convolutional Neural Networks (CNNs), b) Recurrent Neural Networks (RNNs), c) Perceptron, d) Generative Adversarial Networks (GANs).
Convolutional Neural Networks (CNNs)
Q: The Dartmouth Conference in 1956 is known as the birthplace of this particular academic field of computer science?
Artificial Intelligence
