GSCM 410 Final Exam quiz

Quiz

28 questions

About 17 minutes. Every answer comes with the right one.

Quiz

Question types

All 28 questions

Which of the following best describes the primary function of AI as presented in the study material?
Answer
A productivity tool to augment human capabilities
What historical event in 1950 is considered a foundational goal for AI?
Answer
Alan Turing's proposal of the Turing Test
AI is considered a replacement for human intelligence, not an augmentation.
Answer
False
The term 'Artificial Intelligence' was formally coined at the ___ Conference.
Answer
Dartmouth
Which AI approach involves using Artificial Neural Networks (ANNs) with many hidden layers to detect increasingly complex patterns?
Answer
Deep Learning
The Fourth Industrial Revolution is primarily characterized by the integration of which two technologies?
Answer
AI and intelligent automation (Robotics)
Backpropagation uses gradient descent to adjust weights and biases in ANNs.
Answer
True
In LLMs, ___ is the process of breaking down input text into sub-word units.
Answer
tokenization
In a Perceptron, what is the role of 'weights'?
Answer
To represent the strength or importance of each input
What is the primary goal of training an Artificial Neural Network (ANN)?
Answer
To minimize the loss (error) between predicted and actual outputs
The 'Clark-Fisher Hypothesis' suggests that AI will primarily displace manual labor jobs.
Answer
False
___ is a system design that allows an LLM to retrieve external information before generating a response.
Answer
Retrieval Augmented Generation
Which stage of the LLM training pipeline involves exposing the model to a massive corpus of text data using self-supervised learning?
Answer
Pre-training
What is the purpose of 'positional encoding' in the LLM inference flow?
Answer
To preserve the grammatical and contextual order of tokens
A larger context window in an LLM allows it to process more information in a single prompt.
Answer
True
___ refers to the numerical parameters of an AI model that are adjusted during training.
Answer
Model weights
Which component is part of an AI system but not necessarily part of the AI model itself?
Answer
The user interface
What is the function of the 'System Role' when interacting with an LLM via API?
Answer
To provide instructions on how the LLM should behave or adopt a persona
Prompt Injection is a technique used to improve the clarity and structure of prompts.
Answer
False
In Retrieval Augmented Generation (RAG), what is the role of the vector database?
Answer
To store document chunks and their embeddings for fast semantic search
What is the primary benefit of the Mixture of Experts (MoE) architecture?
Answer
It distributes computational load across specialized sub-models for efficiency
Which of the following is a key characteristic of an AI agent's operation?
Answer
It performs multi-step tasks with minimal human intervention
What does 'no-code / low-code solution' aim to achieve in AI development?
Answer
Make AI system development accessible to users without extensive programming knowledge
Chain-of-Thought (CoT) prompting is a technique that:
Answer
Instructs the LLM to reason step-by-step before answering
Explain the concept of 'AI Winters' and their significance in the history of AI.
Answer
AI Winters refer to periods of reduced funding and interest in artificial intelligence research, typically following periods of over-optimism and unmet expectations. These setbacks, occurring in the mid-1970s and after initial neural network challenges, highlight the cyclical nature of AI development and the importance of realistic technological advancement alongside ambitious goals.
Describe the difference between an AI model and an AI system.
Answer
An AI model, such as a Large Language Model (LLM), is the core 'brain' that processes information and generates outputs. An AI system, however, is the complete package that makes the model accessible and useful to users. This includes the model itself, plus essential components like user interfaces, APIs, preprocessing, postprocessing, and coordination logic.
What is 'Model Overfitting' and how can it be avoided during ANN training?
Answer
Model Overfitting occurs when an AI model learns the training data too well, including its noise and specific details, leading to poor performance on new, unseen data. It can be avoided by using techniques such as employing a slower training rate, ensuring the model is exposed to diverse data over multiple epochs, and using regularization methods.
Explain the purpose of the 'gating network' in a Mixture of Experts (MoE) architecture.
Answer
The gating network in an MoE architecture acts as a router. When an input is received, the gating network evaluates it and determines which of the specialized 'expert' sub-models are best suited to process that specific input. It then directs the input to the selected expert(s), ensuring efficient and targeted computation.