Short Answer Questions:

1. What does "generative" mean in the context of AI?

o In AI, "generative" refers to the ability of a model to create new
content, such as text, images, music, or videos, based on
patterns learned from training data.

2. Name two popular applications of generative AI in everyday life.

o Chatbots (e.g., ChatGPT, Google Bard)

o Image generation tools (e.g., DALL·E, MidJourney)

3. What is a Generative Adversarial Network (GAN) and how does it

work?
o A GAN is a type of generative AI model that consists of two
neural networks: a generator that creates new data and a
discriminator that evaluates whether the data is real or fake.
They compete against each other, improving over time to
generate highly realistic content.

4. What is the main difference between a generative AI model and a

discriminative AI model?
o A generative model creates new data similar to its training
data, while a discriminative model classifies data by
distinguishing between different categories.

5. Give one example of how generative AI can be used creatively.

o It can generate original music compositions by learning patterns
from existing songs and creating unique melodies.

6. What is a potential ethical concern related to generative AI?

o The creation of deepfakes, which can be used for
misinformation, identity theft, or spreading false narratives.

Long Answer Questions:

7. Explain how a generative AI model learns to create new content,

using the concept of training data.
o A generative AI model is trained on a large dataset containing
examples of the type of content it needs to generate. It learns
 patterns, structures, and relationships within the data through
techniques like deep learning. During training, the model
adjusts its parameters to minimize the difference between
generated and real data. Over time, it becomes capable of
generating new, original content that resembles the training
data.

8. Discuss the key components of a GAN, including the generator

and discriminator, and how they interact with each other to
produce realistic outputs.
o A GAN consists of two main components:

 Generator: This neural network creates fake data based

on the training dataset. It starts by generating random
noise and refines its output over time.
 Discriminator: This neural network evaluates the
generated data and determines whether it is real (from
the training data) or fake (from the generator).
 Interaction: The generator improves by trying to fool the
discriminator, while the discriminator improves by getting
better at distinguishing real from fake. This adversarial
process continues until the generator produces realistic
outputs that the discriminator cannot easily differentiate
from real data.

9. Describe the potential benefits and drawbacks of using

generative AI in the medical field, providing specific examples.
o Benefits:

 Medical Imaging: AI can generate high-resolution

medical images for training doctors or assisting in
diagnostics (e.g., MRI enhancement).
 Drug Discovery: AI models can generate new molecular
structures, speeding up the development of new
medicines.
o Drawbacks:

 Accuracy Concerns: If trained on biased or incomplete

data, AI-generated medical insights could be inaccurate.
 Ethical Issues: The use of synthetic medical data raises
concerns about privacy and patient consent.
10. Compare and contrast different types of generative AI
models, such as Variational Autoencoders (VAEs) and GANs,
highlighting their strengths and weaknesses in different
scenarios.
o VAEs:

 Strengths: Good for structured data generation (e.g.,

handwriting synthesis, image compression).
 Weaknesses: Can produce blurry images due to
probabilistic nature.
o GANs:

 Strengths: Can create highly realistic images and videos

(e.g., deepfakes, AI art).
 Weaknesses: Difficult to train, may suffer from "mode
collapse" (producing limited variations).
o Comparison: VAEs are better for generating diverse but
slightly less realistic content, while GANs excel in high-quality
realism but may lack variety.

11. Analyze the ethical considerations surrounding deepfakes

created using generative AI, including potential impacts on
society and individuals.
o Misinformation & Fake News: Deepfakes can spread false
information, influencing politics or damaging reputations.
o Privacy Violations: They can be used to create non-
consensual videos, raising concerns about digital identity theft.
o Trust Issues: With deepfakes becoming more realistic, it
becomes harder to verify what is real, leading to societal
distrust in media.
o Regulation & Detection: Ethical AI development should
include deepfake detection systems and legal regulations to
prevent misuse.

12. Imagine you are developing a generative AI model to create

unique artwork. What factors would you need to consider when
designing and training the model to ensure creativity and artistic
quality?
o Training Data: Use a diverse dataset of various artistic styles
to encourage originality.
o Model Type: Consider using GANs or VAEs depending on the
desired balance of creativity and realism.
o Bias & Overfitting: Ensure the model doesn’t just copy
existing artworks but creates new styles.
o Human-AI Collaboration: Allow artists to guide the AI in
refining outputs rather than generating art autonomously.
o Ethical Considerations: Ensure that AI-generated art does not
infringe on copyrights and respects artists' original work.