Generative AI Study Guide

Quiz
Instructions: Answer each question in 2-3 sentences.
. What is the fundamental difference between discriminative AI and generative AI?
9
. Describe how generative adversarial networks (GANs) function.
. Explain the significance of deep learning in the advancement of generative AI.
. What are diffusion models and how do they work in image generation?
8
. Name three examples of generative AI models and give their specific application.
. What is a large language model (LLM) and what are its primary capabilities?
. How does generative AI contribute to the creation of virtual worlds?
7
. What role does the “training” process play in generative AI models?
. How can generative AI be used for data augmentation?
. Give two examples of how generative AI is currently being applied in business or industry.
6
Answer Key
. Discriminative AI learns to distinguish between different classes of data, focusing on
5
classification and pattern recognition. Generative AI, on the other hand, learns to generate
new content similar to the training data, focusing on creating novel outputs.
. GANs operate using two neural networks, a generator that creates new content and a
4
discriminator that tries to distinguish between real and synthetic content. Through this
competition, the generator learns to produce realistic content that can deceive the
discriminator.
3
. Deep learning, which uses neural networks with multiple layers, enabled generative AI to
process complex data patterns, leading to breakthroughs like GANs and VAEs. Deep learning
has significantly improved the quality and complexity of generated content.
. Diffusion models create images by gradually adding noise until the original image is
2
unrecognizable, then reversing the process to remove noise and generate new images. They
are especially effective at producing high-quality images.
. One example is GPT (text generation), another is DALL-E (image generation), and one more is
1
MuseNet (music generation). Each was developed for a specific application using a
generative AI model.
. LLMs are trained on massive text and code datasets, which enable them to understand and
1
0
generate human-like text for various tasks like translation, summarization, and content
9
creation. These models power applications from chatbots to advanced creative writing tools.
8
. Generative AI allows for the creation of highly detailed virtual environments, realistic avatars,
7
and interactive virtual worlds by producing 3D graphics and interactive features. This has
6
applications in metaverse platforms, games, and simulations.
5
. The training process enables AI models to learn from vast datasets. In this process, the
4
models identify patterns and distributions in the data that enable them to make predictions or
3
generate new data.
2
. Generative AI can be used to augment data by producing synthetic data similar to original
1
datasets, thereby increasing dataset size and variability, especially useful when data is
 .

scarce. This leads to more robust and effective machine learning models.
. In business, generative AI is used in customer service to automate tasks via chatbots. Also, AI
1
0
is used to develop marketing materials, and to generate reports, code, and other internal
documents.
9
Essay Questions
Instructions: Choose one of the following essay questions and write a response demonstrating your
understanding of the source material.
. Trace the historical evolution of generative AI, highlighting the key milestones and
breakthroughs that have shaped the field, focusing on the impact of specific advancements
such as deep learning, GANs, and large language models.
. Compare and contrast discriminative AI and generative AI, explaining their distinct
capabilities, typical applications, and how they contribute to the broader field of artificial
intelligence, using examples from the source material.
. Describe the various capabilities of generative AI, including text, image, audio, video, and
code generation. Then analyze the implications and potential applications of each in different
real-world contexts.
. Discuss the ethical considerations that arise from the use of generative AI, focusing on the
potential impact on job markets, the spread of misinformation, and the potential for misuse of
AI-generated content.
. Evaluate the transformative potential of generative AI in various industries, and discuss how it
may change the way humans work, innovate, and interact with technology in the near future.

Glossary of Key Terms

Artificial Intelligence (AI): The simulation of human intelligence in machines, enabling them to
perform tasks that typically require human-like cognitive abilities.
Deep Learning: A subfield of machine learning that utilizes artificial neural networks with multiple
layers to analyze complex patterns in large datasets.
Discriminative AI: An approach to AI that learns to distinguish between different classes or
categories of data, focusing on classification and prediction based on labeled data.
Diffusion Models: Generative models that create images by gradually adding noise to clean images
and then reversing the process to remove noise and generate new images.
Generative Adversarial Networks (GANs): A generative model that uses two neural networks: a
generator that creates new content and a discriminator that attempts to distinguish between real
and synthetic content.
Generative AI: A branch of AI that focuses on developing models to generate new content such as
text, images, audio, and code, based on learning patterns from training data.
Large Language Model (LLM): A type of AI model trained on vast amounts of text data that can
understand, process, and generate human-like text for various applications.
Latent Space: A representation of data that captures the essential features and underlying
structure, enabling the generation of new content by sampling and decoding.
Neural Networks: Computing systems inspired by the biological neural networks of the human
brain, used to process complex data and learn intricate patterns through interconnected nodes or
neurons.
Training: The process of learning from data where AI models adjust their internal parameters to fit
to patterns and structures within the provided dataset.
Transformers: A type of neural network architecture, particularly effective at processing sequential
data such as text, which has led to improvements in language modeling and natural language
processing.
Variational Autoencoders (VAEs): A type of generative model that learns the latent space of
training data, allowing the generation of new data by sampling from this learned representation.