2. Fundamentals of Generative AI Flashcards
(25 cards)
What is a token in the context of generative AI?
A) A security feature
B) A unit of text processed by the model
C) A type of neural network
D) A model evaluation metric
B
In generative AI, a token is a unit of text processed by the model. This is under foundational generative AI concepts.
Which of the following is NOT a typical use case for generative AI models?
A) Image generation
B) Summarization
C) Data encryption
D) Code generation
C
Data encryption is not a typical use case for generative AI models. The other options are potential use cases.
What is the primary advantage of generative AI’s adaptability?
A) It can only work with structured data
B) It can handle a wide range of tasks and domains
C) It always produces perfect results
D) It eliminates the need for human oversight
B
Adaptability in generative AI refers to its ability to handle a wide range of tasks and domains. This is one of the advantages of generative AI.
What is a hallucination in the context of generative AI?
A) A visual output produced by the model
B) A type of model architecture
C) An incorrect or fabricated output presented as fact
D) A method of model training
C
Hallucinations refer to incorrect or fabricated outputs presented as fact by generative AI models. This is listed as a disadvantage of generative AI solutions.
Which AWS service is designed specifically for developing generative AI applications?
A) Amazon EC2
B) Amazon S3
C) Amazon Bedrock
D) Amazon RDS
C
Amazon Bedrock is an AWS service for developing generative AI applications.
What is a foundation model in generative AI?
A) A model that can only generate text
B) A large, pre-trained model that can be adapted for various tasks
C) A model specifically designed for image generation
D) A model that requires no training data
B
A foundation model is a large, pre-trained model that can be adapted for various tasks.
Which of the following is NOT a stage in the foundation model lifecycle?
A) Data selection
B) Pre-training
C) Deployment
D) Marketing
D
Marketing is not a stage in the foundation model lifecycle. The other options are part of the foundation model lifecycle.
What is the primary advantage of using AWS generative AI services for building applications?
A) They are always free
B) They provide a lower barrier to entry
C) They guarantee 100% accuracy
D) They eliminate the need for any coding
B
A lower barrier to entry is one of the advantages of using AWS generative AI services.
What is prompt engineering in the context of generative AI?
A) A method of hardware optimization
B) A technique for designing the physical structure of AI models
C) The process of crafting effective input prompts to guide model outputs
D) A way to reduce energy consumption in AI systems
C
Prompt engineering refers to the process of crafting effective input prompts to guide model outputs.
Which of the following is a potential disadvantage of generative AI solutions?
A) Adaptability
B) Responsiveness
C) Inaccuracy
D) Simplicity
C
Inaccuracy is listed as a potential disadvantage of generative AI solutions.
What is a multi-modal model in generative AI?
A) A model that can only process text data
B) A model that can work with multiple types of data (e.g., text, images, audio)
C) A model that requires multiple GPUs to run
D) A model that can only generate images
B
A multi-modal model can work with multiple types of data.
Which AWS service provides a playground for experimenting with generative AI models?
A) Amazon SageMaker
B) Amazon Comprehend
C) PartyRock
D) Amazon Polly
C
PartyRock, an Amazon Bedrock Playground, is a service for developing generative AI applications.
What is a key consideration when selecting an appropriate generative AI model for a business problem?
A) The model’s popularity on social media
B) The model’s performance requirements
C) The model’s development date
D) The model’s country of origin
B
Performance requirements are one of the factors to consider when selecting appropriate generative AI models.
Which of the following is NOT a typical business metric for evaluating generative AI applications?
A) Conversion rate
B) Average revenue per user
C) Customer lifetime value
D) Model parameter count
D
Model parameter count is not a business metric.
What is a key benefit of AWS infrastructure for generative AI applications?
A) It eliminates the need for any security measures
B) It provides unlimited free computing resources
C) It ensures compliance with relevant regulations
D) It guarantees that AI models will never make mistakes
C
Compliance is one of the benefits of AWS infrastructure for generative AI applications.
What is chunking in the context of generative AI?
A) A method of data compression
B) A technique for breaking down large inputs into smaller, manageable pieces
C) A type of model architecture
D) A way to increase model accuracy
B
Chunking refers to breaking down large inputs into smaller, manageable pieces.
Which of the following is a key advantage of generative AI’s simplicity? A) It always produces perfect results
B) It requires no human input
C) It can be easier to implement and use compared to traditional methods
D) It eliminates the need for data preprocessing
C
Simplicity in generative AI often means it can be easier to implement and use compared to traditional methods.
What is a diffusion model in generative AI?
A) A model that only works with textual data
B) A type of generative model often used for image generation
C) A model that requires no training data
D) A model specifically designed for natural language processing
B
Diffusion models are a type of generative model often used for image generation.
Which AWS service is designed to provide conversational AI capabilities?
A) Amazon Bedrock
B) Amazon SageMaker
C) Amazon Q
D) Amazon S3
C
Amazon Q is an AWS service for developing generative AI applications, and it provides conversational AI capabilities.
What is a key consideration in the cost tradeoffs of AWS generative AI services?
A) The color scheme of the user interface
B) The number of employees in the company
C) Token-based pricing
D) The physical location of the data center
C
Token-based pricing is one of the cost tradeoffs to consider for AWS generative AI services.
What is the primary purpose of embeddings in generative AI?
A) To compress data for storage
B) To represent data in a high-dimensional space
C) To encrypt sensitive information
D) To generate random numbers
B
Embeddings are used to represent data in a high-dimensional space.
Which of the following is NOT a typical use case for generative AI in customer service?
A) Chatbots
B) Automated email responses
C) Physical robot assistants
D) FAQ generation
C
Physical robot assistants are not a typical use case for generative AI in customer service. The other options align with the use cases.
What is a key advantage of using AWS generative AI services for building applications in terms of development speed?
A) They automatically write all the code for you
B) They provide faster time to market
C) They eliminate the need for testing
D) They guarantee instant deployment
B
Speed to market is one of the advantages of using AWS generative AI services.
What is nondeterminism in the context of generative AI?
A) A type of model architecture
B) A method of data preprocessing
C) The property of producing different outputs for the same input
D) A technique for improving model accuracy
C
Nondeterminism refers to the property of producing different outputs for the same input. This is listed as a potential disadvantage of generative.
