University of Engineering and Technology, Taxila

Software Engineering Department

Information & Communication Technologies (SE-101)

Assignment No.2
Topic: Generative AI
Submitted to: Dr. Kanwal Yousaf
Date of submission: 26th September, 2023
Section: Alpha
Submitted by
Alina Nisar(23-SE-03)
Kiran Fatima(23-SE-05)
Ifrah Sarfraz(23-SE-09)
Aneesa Ashfaq(23-SE-13)
Onaiza Maryam(23-SE-87)
Contents
Introduction [Kiran Fatima(23-SE-05)] ..................................................................................... 4
Who created generative AI? ........................................................................................................ 4
Transformers and large language models (LLMs) ...................................................................... 4
Transformers ............................................................................................................................... 4
How neural networks are transforming generative AI? .............................................................. 6
Popular generative AI interfaces ................................................................................................. 6
Dall-E. ..................................................................................................................................... 6
ChatGPT. ................................................................................................................................. 6
Bard ......................................................................................................................................... 7
How could generative AI replace jobs? ...................................................................................... 7
Ethics and bias in generative AI.................................................................................................. 7
Generative AI history .................................................................................................................. 8
Best practices for using generative AI ........................................................................................ 9
Working of Generative AI [Ifrah Sarfraz(23-SE-09)] ............................................................... 9
Start with the brain .................................................................................................................... 10
Build an artificial neural network ............................................................................................. 10
Teach the newborn network model ............................................................................................11
Role of machine learning in generative AI ............................................................................... 12
Applications of Generative AI [Alina Nisar(23-SE-03)] .......................................................... 13
Text Generation:........................................................................................................................ 13
Explanation: .......................................................................................................................... 14
Examples of Real-World Text-generation Applications: ...................................................... 15
Top AI Text Generators of 2023............................................................................................ 15
Image Generation: ..................................................................................................................... 16
How do they work? ............................................................................................................... 16
Use Cases and Applications of Image Generators ................................................................ 17
Top AI Image Generators of 2023 ........................................................................................ 17
Style Transfer AI ....................................................................................................................... 17
How do they work? ............................................................................................................... 18
 Applications and Use Cases .................................................................................................. 18
Top Style transfer tools ......................................................................................................... 19
Data Augmentation: .................................................................................................................. 19
How does it work: ................................................................................................................. 20
Applications : ........................................................................................................................ 20
Data Augmentation tools and libraries:................................................................................. 21
Generative AI Tools [Aneesa Ashfaq (23-SE-13)] .................................................................... 21
1. AI Art Generator – Midjourney: ........................................................................................... 21
2. Google Bard: ......................................................................................................................... 23
How does Google Bard work? .............................................................................................. 23
3. Chat GPT ............................................................................................................................. 24
4. CONSENSUS.AI .................................................................................................................. 28
Benefits of ConsensusAI? ..................................................................................................... 28
5. Glasp ..................................................................................................................................... 29
Pros and cons of Generative AI [Onaiza Maryam(23-SE-87)] ............................................... 30
Benefits of Generative AI ......................................................................................................... 30
1. Automates and innovates content generation:................................................................... 30
2. Enhances customer experience through personalization: ................................................. 31
3. Optimizes product designs: ............................................................................................... 31
4. Strengthens cybersecurity efforts: ..................................................................................... 31
5. Advances health care research: ......................................................................................... 32
6. Streamlines business processes: ........................................................................................ 32
7. Improved customer service: .............................................................................................. 33
8. Fosters market innovation:................................................................................................ 33
9. Drives digital transformation: ........................................................................................... 34
10. Inspires creativity: ........................................................................................................... 34
Disadvantages of Generative AI ............................................................................................... 34
Future of Generative AI [Ifrah Sarfraz(23-SE-09)]…………………………………………..35
Conclusion…………………………………………………………………………………...….39
 Introduction [Kiran Fatima(23-SE-05)]1
Generative AI is a type of artificial intelligence technology that can produce various types of
content including text, images, audio and synthetic data. The recent buzz around generative AI has
been driven by the simplicity of new user interfaces for creating high-quality text, graphics and
videos in a matter of seconds. Generative AI focuses on creating new and original content, chat
responses, designs, synthetic data or even deepfakes. It's particularly valuable in creative fields
and for novel problem-solving, as it can autonomously generate many types of new outputs.

The technology, it should be noted, is not brand-new. Generative AI was introduced in the 1960s
in chatbots. But it was not until 2014, with the introduction of generative adversarial networks, or
GANs (a type of machine learning algorithm) that generative AI could create convincingly
authentic images, videos and audio of real people.

On one hand, this newfound capability has opened up opportunity that include better movie
dubbing and rich educational content. It also unlocked concerns about deepfakes (digitally forged
images or videos) and harmful cybersecurity attacks on businesses including nefarious requests
that realistically mimic an employee’s boss.

Who created generative AI?

Joseph Weizenbaum created the first generative AI in the 1960s as part of the Eliza chatbot.

Ian Goodfellow demonstrated generative adversarial networks for generating realistic-looking and
-sounding people in 2014.

Subsequent research into LLMs from Open AI and Google ignited the recent enthusiasm that has
evolved into tools like ChatGPT, Google Bard and Dall-E.

Transformers and large language models (LLMs)

Two additional recent advances that have played a critical part in generative AI going in
mainstream are transformers and large language models they enabled.

Transformers are a type of machine learning that made it possible for researchers to train ever-
larger models without having to label all of the data in advance. New models could thus be trained

1
https://www.techtarget.com/searchenterpriseai/definition/generative-AI
on billions of pages of text, resulting in answers with more depth. In addition, transformers
unlocked a new notion called attention that enabled models to track the connections between words
across pages, chapters and books rather than just in individual sentences. And not just words:
Transformers could also use their ability to track connections to analyze code, proteins, chemicals
and DNA.

The rapid advances in so-called large language models (LLMs) -- i.e., models with billions or even
trillions of parameters -- have opened a new era in which generative AI models can write engaging
text, paint photorealistic images and even create somewhat entertaining sitcoms on the fly.
Moreover, innovations in multimodal AI enable teams to generate content across multiple types of
media, including text, graphics and video. This is the basis for tools like Dall-E that automatically
create images from a text description or generate text captions from images.

These breakthroughs notwithstanding, we are still in the early days of using generative AI to create
readable text and photorealistic stylized graphics. Early implementations have had issues with
accuracy and bias, as well as being prone to hallucinations and spitting back weird answers.
Still, progress thus far indicates that the inherent capabilities of this generative AI could
fundamentally change enterprise technology how businesses operate. Going forward, this
technology could help write code, design new drugs, develop products, redesign business
processes and transform supply chains.

Generative AI starts with a prompt that could be in the form of a text, an image, a video, a design,
musical notes, or any input that the AI system can process. Various AI algorithms then return new
content in response to the prompt. Content can include essays, solutions to problems, or realistic
fakes created from pictures or audio of a person.

Now, pioneers in generative AI are developing better user experiences that let you describe a
request in plain language. After an initial response, you can also customize the results with
feedback about the style, tone and other elements you want the generated content to reflect.
How neural networks are transforming generative AI?
Researchers have been creating AI and other tools for programmatically generating content since
the early days of AI. The earliest approaches, known as rule-based systems and later as "expert
systems," used explicitly crafted rules for generating responses or data sets.

Neural networks, which form the basis of much of the AI and machine learning applications today,
flipped the problem around. Designed to mimic how the human brain works, neural networks
"learn" the rules from finding patterns in existing data sets. Developed in the 1950s and 1960s, the
first neural networks were limited by a lack of computational power and small data sets. It was not
until the advent of big data in the mid-2000s and improvements in computer hardware that neural
networks became practical for generating content.

The field accelerated when researchers found a way to get neural networks to run in parallel across
the graphics processing units (GPUs) that were being used in the computer gaming industry to
render video games. New machine learning techniques developed in the past decade, including the
aforementioned generative adversarial networks and transformers, have set the stage for the recent
remarkable advances in AI-generated content.

Popular generative AI interfaces

Dall-E. Trained on a large data set of images and their associated text descriptions, Dall-E is an
example of a multimodal AI application that identifies connections across multiple media, such as
vision, text and audio. In this case, it connects the meaning of words to visual elements. It was
built using OpenAI's GPT implementation in 2021. Dall-E 2, a second, more capable version, was
released in 2022. It enables users to generate imagery in multiple styles driven by user prompts.

ChatGPT. The AI-powered chatbot that took the world by storm in November 2022 was built on
OpenAI's GPT-3.5 implementation. OpenAI has provided a way to interact and fine-tune text
responses via a chat interface with interactive feedback. Earlier versions of GPT were only
accessible via an API. GPT-4 was released March 14, 2023. ChatGPT incorporates the history of
its conversation with a user into its results, simulating a real conversation. After the incredible
popularity of the new GPT interface, Microsoft announced a significant new investment into
OpenAI and integrated a version of GPT into its Bing search engine.

Bard. Google was another early leader in pioneering transformer AI techniques for processing
language, proteins and other types of content. It open sourced some of these models for researchers.
However, it never released a public interface for these models. Microsoft's decision to implement
GPT into Bing drove Google to rush to market a public-facing chatbot, Google Bard, built on a
lightweight version of its LaMDA family of large language models. Google suffered a significant
loss in stock price following Bard's rushed debut after the language model incorrectly said the
Webb telescope was the first to discover a planet in a foreign solar system. Meanwhile, Microsoft
and ChatGPT implementations also lost face in their early outings due to inaccurate results and
erratic behavior. Google has since unveiled a new version of Bard built on its most advanced
LLM, PaLM 2, which allows Bard to be more efficient and visual in its response to user queries.

How could generative AI replace jobs?

Generative AI has the potential to replace a variety of jobs, including the following:

 Writing product descriptions.

 Creating marketing copy.
 Generating basic web content.
 Initiating interactive sales outreach.
 Answering customer questions.
 Making graphics for webpages.

Some companies will look for opportunities to replace humans where possible, while others will
use generative AI to augment and enhance their existing workforce.

Ethics and bias in generative AI

Despite their promise, the new generative AI tools open a can of worms regarding accuracy,
trustworthiness, bias, hallucination and plagiarism -- ethical issues that likely will take years to
sort out. None of the issues are particularly new to AI. Microsoft's first foray into chatbots in 2016,
called Tay, for example, had to be turned off after it started spewing inflammatory rhetoric on
Twitter.
What is new is that the latest crop of generative AI apps sounds more coherent on the surface. But
this combination of humanlike language and coherence is not synonymous with human
intelligence, and there currently is great debate about whether generative AI models can be trained
to have reasoning ability. One Google engineer was even fired after publicly declaring the
company's generative AI app, Language Models for Dialog Applications (LaMDA), was sentient.

The convincing realism of generative AI content introduces a new set of AI risks. It makes it harder
to detect AI-generated content and, more importantly, makes it more difficult to detect when things
are wrong. This can be a big problem when we rely on generative AI results to write code or
provide medical advice. Many results of generative AI are not transparent, so it is hard to determine
if, for example, they infringe on copyrights or if there is problem with the original sources from
which they draw results. If you don't know how the AI came to a conclusion, you cannot reason
about why it might be wrong.

Generative AI history
The Eliza chatbot created by Joseph Weizenbaum in the 1960s was one of the earliest examples of
generative AI. These early implementations used a rules-based approach that broke easily due to a
limited vocabulary, lack of context and overreliance on patterns, among other shortcomings. Early
chatbots were also difficult to customize and extend.

The field saw a resurgence in the wake of advances in neural networks and deep learning in 2010
that enabled the technology to automatically learn to parse existing text, classify image elements
and transcribe audio.

Ian Goodfellow introduced GANs in 2014. This deep learning technique provided a novel
approach for organizing competing neural networks to generate and then rate content variations.
These could generate realistic people, voices, music and text. This inspired interest in -- and fear
of -- how generative AI could be used to create realistic deepfakes that impersonate voices and
people in videos.

Since then, progress in other neural network techniques and architectures has helped expand
generative AI capabilities. Techniques include VAEs, long short-term memory, transformers,
diffusion models and neural radiance fields.
Best practices for using generative AI
The best practices for using generative AI will vary depending on the modalities, workflow and
desired goals. That said, it is important to consider essential factors such as accuracy, transparency
and ease of use in working with generative AI. The following practices help achieve these factors:

 Clearly label all generative AI content for users and consumers.

 Vet the accuracy of generated content using primary sources where applicable.
 Consider how bias might get woven into generated AI results.
 Double-check the quality of AI-generated code and content using other tools.
 Learn the strengths and limitations of each generative AI tool.
 Familiarize yourself with common failure modes in results and work around these.

Working of Generative AI [Ifrah Sarfraz(23-SE-09)]

Generative AI is a fascinating field that uses various techniques, like neural networks that are
mathematical models and deep learning algorithms, to identify patterns and generate new
outcomes based on them. It creates a replica of the human brain to understand the structures and
patterns of the data. Another advantage is that you can train neural networks using various learning
methods, such as unsupervised or semi-supervised. This help organizations build foundation
models and general-purpose AI models that can perform multiple tasks. Some example of
foundation models are GPT-3 and Stable Diffusion, which are based on natural language
processing. They can generate text or image from a given input.

“We don’t know how they do the actual creative task because what goes on inside the neural
network layers is way too complex for us to decipher, at least today;”2

The training process for a generative model involves feeding it a large dataset of examples, such
as image, text, audio, and videos. Then, the model analyzes the patterns and relationship within
the input data to understand the underlying rules governing the content. It generate new data by
sampling from probability distribution it has learned. And it continuously adjusts its parameters to
maximize the probability of generating accurate output.

2
Dean Thompson , a former chief technology officer of multiple AI startups that have been acquired over years by
companies
Here’s an overview of what we do know about how generative AI works:

Start with the brain

A good place to start in understanding generative AI models is with the human brain3, says Jeff
Hawkins. Hawkins hypothesized that, at the neuron level, the brain works by continuously
predicting what’s going to happen next and then learning from the difference between its
predictions and subsequent reality. To improve its predictive ability, the brain builds an internal
representation of the world. In his theory, human intelligence emerges from that process. Whether
influenced by Hawkins or not, generative AI works exactly this way. And, startlingly, it acts as if
it is intelligent.

Build an artificial neural network

All generative AI models begin with an artificial neural network encoded in software. Thompson
says a good visual metaphor for a neural network is to imagine the familiar spreadsheet, but in
three dimensions because the artificial neuron are stacked in layers, similar to how real neurons
are stacked in the brain. AI researchers even call each neuron a “cell”, Thompson notes, and each
cell contains a formula relating it to other cells in the network---mimicking the way that the
connections between brain neuron have different strengths.

3
Jeff Hawkins in his book “On Intelligence”, 2004
Each layer may have tens, hundreds, or thousands of artificial neurons, but the number of number
of neurons is not what AI researchers focus on. Instead, they measure models by the number of
connections between neurons. The strengths of these connections vary based on their cell
equations’ coefficients, which are more generally called “weights” or “parameters”. These
connection-defining coefficients are what’s being referred to when you read, for example, that the
GPT-3 model has 175 billion parameters. The latest version, GPT-4, is rumored to have trillions of
parameters, though that is unconfirmed. There are a handful of neural network architectures with
differing characteristics that lend themselves to producing content in a particular modality; the
transformer architecture appears to be best for large language models, for example.

Teach the newborn network model

Large language models are given enormous volumes of text to process and tasked to make simple
predictions, such as the next word in sequence or the correct order of a set sentences. In practice,
though, neural network models work in units called tokens, not words.

“A common word may have its own token, uncommon word would certified be made up of
multiple token, and some token may just be single space followed by ‘th’ because that sequence
of three characters is so common,” said Thompson. To make each prediction, the model inputs a
token at the bottom layer of a particular stack of artificial neurons; that layer processes it and passes
its output emerges from the top of the stack. Stack sizes can vary significantly, but they’re generally
on the order of tens of layers, not thousands or millions.

In the early training stages, the model’s prediction aren’t very good. But each time the model
predicts a token, it checks for correctness against the training data. Whether it’s right or wrong, a
“backpropagation” algorithm adjusts the parameters---that is, the formulas’ coefficients---in each
cell of the stack that made that prediction. The goal of the adjustments is to make the correct
prediction more probable.

"It does this for right answers, too, because that right prediction may have only had, say, a 30%
certainty, but that 30% was the most of all the other possible answers," Thompson said. "So,
backpropagation seeks to turn that 30% into 30.001%, or something like that."4

4
https://www.oracle.com
After the model has repeated this process for trillions of text tokens, it becomes very good at
predicting the next token, or word. After initial training, generative Al models can be fine-tuned
via a supervised learning technique, such as reinforcement learning from human feedback (RLHF).
In RLHF, the model's output is given to human reviewers who make a binary positive or negative
assessment-thumbs up or down-which is fed back to the model. RLHF was used to fine-tune
OpenAl's GPT 3.5 model to help create the ChatGPT chatbot that went viral.

Role of machine learning in generative AI

Generative AI falls under machine learning and is capable of crafting fresh content resembling
what already exists. We educate models to fashion items akin to those they've encountered earlier.

Subsequently, these models employ their acquired knowledge to produce novel content akin to the
examples. These models put their developed understanding to work by creating unknown content
resembling the given criteria.

This process is facilitated through various methods, including utilizing such as Generative
Adversarial Networks (GANs) and Variational Autoencoders These tools employ machine learning
to generate new content mirroring established patterns.

Machine learning helps it to make new things that follow patterns from the old stuff. It's like
copying a style to create Networks (GANs) and Variational Autoencoders (VAEs). These tools
employ machine to generate new content mirroring established patterns.

Machine learning helps it to make new things that follow patterns from the old stuff. It's like
copying a style to create new pictures, or words.
Applications of Generative AI [Alina Nisar(23-SE-03)]
Text Generation:
This technology has the potential to completely transform communication in a world where
words have the ability to inspire, encourage, and educate. As advancements in Artificial
Intelligence (AI) continue to push the boundaries of what is possible, text generation stands at the
forefront of this digital evolution.

“Text generation is a process where an AI system produces written content, imitating human
language patterns and styles. The process involves generating coherent and meaningful text that
resembles natural human communication.”

The potential for AI to produce text that looks like human writing is enormous, and it might
revolutionize a variety of professions, including journalism, content development, customer
service, and personal contact.
Explanation:
Algorithms and language models are used in text creation to process input data and produce text
output. In order to teach AI models patterns, syntax, and contextual information, massive text
datasets are used for training. These models then create new text in response to cues or
requirements by applying the knowledge they have acquired.

Language models, like Google's PaLM and GPT (Generative Pre-trained Transformer), which have
been trained on enormous volumes of text data from the internet, are the foundation of text
production. These models use deep learning methods, particularly neural networks, to comprehend
sentence structure and produce content that is coherent and appropriate for the context. The AI
model employs a seed input, like a sentence or keyword, to generate text during the process. 5

https://www.datacamp.com/blog/what-is-text-generation
6
https://huggingface.co/tasks/text-generation
Examples of Real-World Text-generation Applications:
Text generation has applications in many real-world life scenarios:

 Content creation: AI-powered systems can generate articles, blog posts, and product
descriptions. These systems are trained on vast amounts of data and can produce coherent
content in a fraction of the time it would take a human writer.
 Chatbots and virtual assistants. AI-powered chatbots and virtual assistants are able to
comprehend customer inquiries and respond with pertinent information and support that is
tailored to the individual.
 Language translation. Language translation services can be enhanced by using text-
generation models. Artificial intelligence (AI) models can provide precise translations
instantly by examining vast amounts of translated text. This improves communication
between speakers of different languages.
 Summarization. Text summarization offers a concise version of information by
identifying the most important points, which can help generate summaries of research
papers, blog posts, news articles, chapters, and books.

Top AI Text Generators of 2023:

1. Click-up AI: Besides being embedded inside Docs and tasks, what sets ClickUp AI apart
are the 100+ tools that use prompts tailored to specific roles.
2. QuillBot: It describes itself as a paraphrasing tool rather than an AI text generator.
3. Jasper: Jasper makes this list of AI text generators because it’s one of the best that focuses
on reflecting the user’s brand style and voice.
4. Copy.ai: It is ideal for marketing teams to create sales and marketing copy.
5. ContentBot.ai: This AI writing tool is best known for writing long-form blog posts but it
can also paraphrase content and generate AI copywriting.
6. ChatGPT: It uses the latest OpenAI GPT-3.5 large language model (LLM) for the free
version and the GPT-4 LLM for the paid ChatGPT Plus.
 7. Rytr: Rytr stands out for supporting over 30 languages and owns a plagiarism checker, SEO
tool, and image creation option.7

Image Generation:

Visual effects (VFX) and computer-generated imagery (CGI), the technologies that have made it
simpler to produce realistic images that heighten immersion, are no strangers to consumers today.
These cutting-edge technologies have allowed us to appreciate some of our favorite films, like
Jurassic Park and Avatar. Although these animation tools have been available for a while, AI Image
Generators are a new technology that is causing quite a stir in the content industry.

“AI image generators utilize trained artificial neural networks to create images from scratch.
These generators have the capacity to create original, realistic visuals based on textual input
provided in natural language.”8

How do they work?

A sophisticated AI image generator utilizes an advanced machine learning algorithm called
artificial neural networks (ANN), inspired by the structure of biological neural networks. The ANN
is trained on a vast dataset containing pairs of images and corresponding text. Through this training
process, the AI system comprehensively learns various elements, including the styles of artists like
Van Gogh, the nuances of colors such as blue, and the visual characteristics of diverse objects.

7
https://clickup.com/blog/ai-text-generator/#33-7-rytr
8
https://www.altexsoft.com/blog/ai-image-generation/
9
https://www.researchgate.net/figure/Images-generation-types-based-on-the-type-of-the-input_fig1_359177684
Once trained, the AI image generator can rapidly interpret any given text prompt, generating new
images that emulate various art forms almost instantly for the user to observe.

Use Cases and Applications of Image Generators:

 Entertainment: In the entertainment industry, AI image generators create realistic

environments and characters for video games and movies. This saves time and resources
that would be used to manually create these elements. Examples are The Frost and Dall-E
2.
 Marketing and advertising: Within the realm of marketing and advertising, AI-generated
images efficiently create visuals for campaigns. Rather than arranging a photoshoot for a
new product, marketers can leverage AI to produce top-notch images suitable for
promotional materials at a swift pace.
 Medical Imaging: In the medical domain, AI image generators are pivotal in enhancing
the quality of diagnostic images. For instance, AI can be employed to produce more precise
and detailed images of tissues and organs, contributing to more accurate diagnostic
processes.10

Top AI Image Generators of 2023:

 DreamStudio (Stable Diffusion): Impressive image quality with a user-friendly interface.

 Midjourney: Advanced image generation AI wizards.
 DALL-E 2: The Classic Tool for Digital Art (DALL-E 3 Upcoming).
 Fotor: The most simple tool with good image quality.
 Craiyon: A Free Tool for Simple AI Image Generation.
 Adobe Firefly: A New Tool with a Great Potential.11

Style Transfer AI:

“Style transfer is a computer vision technique that takes two images—a content image and a style
reference image—and blends them together so that the resulting output image retains the core

10
https://narrato.io/blog/what-are-ai-image-generators-how-do-they-work/
https://www.semrush.com/goodcontent/content-marketing-blog/best-ai-image-
11

generator/#3_dall_e_2_the_classic_tool_for_digital_art_dall_e_3_upcoming
elements of the content image, but appears to be “painted” in the style of the style reference
image.”

In traditional supervised learning for style transfer, a pair of images is required—a base image and
its artistic representation. However, this approach is often impractical due to the scarcity of such
image pairs.

How do they work?

Style transfer utilizes pre-trained convolutional neural networks (CNNs) to extract content and
style representations from input images. By minimizing a loss function that balances content and
style differences, the algorithm iteratively adjusts the pixel values of a generated image to achieve
the desired fusion of style and content.

Applications and Use Cases:

Style Transfer can be used in the following areas:

 Photo and video editor: Style transfer software enhances photo and video editing by
adding popular art styles. It's easy to incorporate on mobile phones due to deep learning
techniques' flexibility, allowing real-time processing and transformation. This makes
access to high-quality editing tools more accessible than ever before.
 Commercial art: AI-powered artwork, including poetry, art, and music, is transforming
our understanding of uniqueness and presentation. Style transfer could be used to create
high-quality prints for large-scale advertising campaigns or office buildings, potentially
altering our perception of the effects of art on commerce.
 Gaming: Google unveiled Stadia, a cloud-based video game streaming service, at the 2019
Game Developers Conference. The service features an in-game style transfer function that
automatically rearranges the virtual environment using textures and color schemes from
various artistic genres. Google aims to empower the artist within every developer, making
artistic creativity more accessible.
  Virtual Reality: VR style transfer applications are in the research stage, but Facebook
claims it has the potential to fundamentally change how VR creators create visual
narratives for games, apps, movies, and other media, similar to gaming.12

Top Style transfer tools:

The 4 top-rated style transfer tools are as follows:

 Style Transfer generators: Style transfer generators revolutionize digital artistry by

allowing users to effortlessly overlay renowned art styles onto various images.
 GAN (Generative Adversarial Network) Generators: use dueling neural networks to
create ultra-realistic images, offering artists and researchers opportunities and showcasing
advancements in AI technology.
 Deep Dream Generators: are AI image generators that create surreal, bizarre artworks by
identifying and enhancing patterns in images, offering a deep exploration of dreams
synthesized through neural networks.
 Text-to-image generator: are revolutionizing storytelling by transforming text into vivid
visuals, bridging the gap between words and visuals. They create a harmonious relationship
between narratives and imagery, enhancing conceptual design processes.13

Data Augmentation:

“Data augmentation is a set of techniques to artificially increase the amount of data by generating
new data points from existing data. This includes making small changes to data or using deep lear
ning models to generate new data points.”

Data augmentation is a valuable technique that enhances machine learning models' performance
and results by creating novel and distinct training datasets. A machine learning model works better
and more accurately if the dataset is rich and sufficient.

12
https://fritz.ai/style-transfer/#:~:text=Introduction,turns%20this%20into%20a%20reality.
13
https://techvify-software.com/best-ai-image-generators/
The methods of gathering and labelling data for machine learning models can be time-consuming
and expensive. Employing data augmentation techniques to transform datasets enables businesses
to lower these operating expenses. Cleaning data is crucial for high-accuracy models, but reducing
data representability can hinder accurate predictions. Data augmentation approaches can enhance
machine learning models' resilience by creating real-world variations.14

How does it work:

In data augmentation, various techniques are employed to enhance the diversity of training datasets
across different domains. In the context of audio data, noise injection, shifting, speed alteration,
and pitch modification are common practices to improve model performance. For text data,
strategies like word or sentence shuffling, word replacement with synonyms, syntax-tree
manipulation, random word insertion, and deletion are employed to create variations. Image
augmentation involves geometric transformations, color space manipulations, kernel filters,
random erasing, and mixing of images, but caution is advised to avoid negatively impacting model
performance when applying multiple transformations. Advanced techniques include the use of
Generative Adversarial Networks (GANs) for generating synthetic data points without relying on
existing data and Neural Style Transfer, which employs convolutional layers to dissect images into
context and style components. Mastery of these techniques contributes to robust model training
and improved generalization to diverse data scenarios.

Applications :

 Healthcare:

Medical imaging dataset acquisition and labelling are costly and time-consuming, requiring
subject matter validation. Machine learning models can be trained using geometric
transformations, such as random cropping, zooming, stretching, and color space change, but should
be used cautiously.

 Self-driving cars:

14
https://research.aimultiple.com/data-augmentation/
Companies are utilizing simulated environments to generate synthetic data using reinforcement
learning for self-driving car training and testing, enhancing data security and generating real-world
scenarios.

 Natural language processing:

Text data augmentation is utilized for improving performance in low-resource languages, high-
error recognition scenarios, sequence-to-sequence data generation, and text classification,
particularly in Automatic Speech Recognition.15

Data Augmentation tools and libraries:

Some of the most popular and powerful data augmentation tools and libraries are:

1. Augmentor: An extensive range of transformations, pipelines, and operations are

supported by this Python module for picture augmentation.
2. NLP-Aug: A Python library for text and audio augmentation that supports various natural
language processing and speech processing techniques.
3. Albumentation: A Python library for image augmentation that is fast, flexible, and
compatible with PyTorch, TensorFlow, Keras, and other frameworks.

4. Imgaug: A Python library designed for image augmentation, catering to a broad range of
transformations, including both basic and intricate changes such as geometric alterations,
color manipulations, blurring, and the addition of weather effects.16

Generative AI Tools [Aneesa Ashfaq (23-SE-13)]

1. AI Art Generator – Midjourney:

Midjourney is an artificial intelligence bot that functions through machine learning to form
visuals from written text / prompts. Essentially, it is an autonomous program operating on the net

15
https://www.datacamp.com/tutorial/complete-guide-data-augmentation
16
https://www.linkedin.com/advice/1/what-best-data-augmentation-techniques
or other network capable of conversing with customers or systems. It seems to turn thoughts into
realities.

Several similar technologies have emerged including DALL-E 2 and MidJourney. Describe
Midjourney AI what image you would like to see. It goes out and collects pictures to understand
what has been articulated so far; and executes a distinctive and creative combination as the final
output to put in front of you. 17

Midjourney gives you some kind of a free trial. You can create your first 25 images on Midjourney
for free. The monthly fee for the Basic Membership gives you the opportunity to make a maximum
of 200 pictures, while subscribing to Standard or Pro Membership gives you access to unlimited
number of creations and more concurrent jobs.

PROMPT 1:

Astronaut walking on the red dusty surface of Mars, collecting samples of its soil, rocks, and
atmosphere. With every step, they are discovering new and exciting things about this alien world.

17
https://textcortex.com/post/chatgpt-prompts-for-midjourney
PROMPT 2:

Dragon, flying across the world. Its wings were so large that they cast a shadow over the land,
blocking out the sun. Its fire was so powerful that it could turn entire cities to ash in a matter of
moments.

2. Google Bard:
Google Bard is an AI-powered chatbot tool designed by Google to simulate human
conversations using natural language processing and machine learning. In addition to
supplementing Google search, Bard can be integrated into websites, messaging platforms or
applications to provide realistic, natural language responses to user questions. 18

How does Google Bard work?

Google Bard is built on the Pathways Language Model 2 (PaLM 2), a language model released in
late 2022.

PaLM and the model that preceded it, Google's Language Model for Dialogue Applications
(LaMDA) technology, are based on Transformers, Google's neural network architecture released

18
https://www.techtarget.com/searchenterpriseai/definition/Google-Bard
in 2017. Because Google released the Transformer architecture as open source, it has been the
framework for other generative AI tools, including the GPT-3 language model used in ChatGPT.

Bard is designed around search. It aims to allow for more natural language queries, rather than
keywords, for search. Bard's AI is trained around natural-sounding conversational queries and
responses. Instead of just giving a list of answers, it provides context to the responses. It is also
designed to help with follow-up questions -- something new to search. Bard has a share
conversation function and a double check function that helps users fact-check generated results.

Bard can also access information from a number of Google apps and services, including YouTube,
Maps, Hotels, Flights, Gmail, Docs and Drive, letting users apply Bard to their personal content.

PROMPT: What makes a Password strong?

3. Chat GPT:
Software called an ai code generator python uses machine learning methods to create code
automatically. These tools help programmers write code more quickly, reducing time spent on
boring jobs and boosting output in general. Python, Java, and HTML are just a few computer
languages where AI code generator can provide code samples.

ChatGPT is an AI language model that can produce code fragments in response to questions in
natural language. It uses OpenAI's sophisticated language processing capabilities to help
programmers write code more quickly.

Highlights:

 Flexibility: This AI chatbot is far more versatile than just coding. It might only be a
question-and-answer tool, but it can streamline workflows for every function in your
company.
 Follow-up inquiries: Some AI code generators fall short right away because they don't
comprehend the whole context of your query. ChatGPT can assist you in determining the
details you need to supply to receive an accurate result since it poses follow-up questions.
 Code evaluation: ChatGPT is particularly skilled at finding flaws in your code and assisting
you in fixing them.

PROMPT 1: Write a code that defines four functions for addition, subtraction, multiplication, and
division.

Here is a simple calculator code in Python:

def add(x, y):

return x + y

def subtract(x, y):

return x - y

def multiply(x, y):

return x * y

def divide(x, y):

return x / y
print("Select operation.")
print("1.Add")
print("2.Subtract")
print("3.Multiply")
print("4.Divide")

while True:
choice = input("Enter choice(1/2/3/4): ")

if choice in ('1', '2', '3', '4'):

num1 = float(input("Enter first number: "))
num2 = float(input("Enter second number: "))

if choice == '1':
print(num1, "+", num2, "=", add(num1, num2))

elif choice == '2':

print(num1, "-", num2, "=", subtract(num1, num2))

elif choice == '3':

print(num1, "*", num2, "=", multiply(num1, num2))

elif choice == '4':

print(num1, "/", num2, "=", divide(num1, num2))
break
else:
print("Invalid Input")
PROMPT 2: Tell me all a software that can check and find errors in my writing.

PROMPT 3: You have heared the story of heer ranjha . At the end both died . Tell me the an
alternative end of this story.
4. CONSENSUS.AI :

Consensus is a search engine. It uses AI to find research pieces for scientific study. Consensus
AI provides evidence-based answers faster. It aims to make expert knowledge accessible and
consumable for all. So that you don’t have to struggle to find authentic sources and collect
information after a lot of hard work.

Benefits of ConsensusAI?
This AI tool for scientific research offers many benefits. Here are important ones for you to know.

There is no ad on this search engine. So, users can navigate smoothly without any interruptions. It
leads to faster loading and improves the page experience. You can find scientific research without
any distractions.The consensus app provides reliable and fact-checking information to you. It uses
a network of AI algorithms to cross-reference and confirm details from numerous sources.

You can think of Consensus AI as a super-smart search assistant. The AI-powered app pulls
information from multiple sources simultaneously. Gone are the days when you have to use a pool
of books and dive deep from it to search for reliable, relatable, factual, and authentic information
for your research paper. Now it takes only a few seconds with the help of the Consensus app.

Consensus AI is naturally aligned with good SEO practices. It provides quality content, reliable
and accurate information. This type of content naturally attracts readers. Moreover, the tool verifies
information from multiple sources, so the content you get is more reliable. 19

19
https://www.technource.com/blog/what-is-consensus-ai/
PROMPT: Do direct cash transfers reduces poverty?

5. Glasp:

Glasp is a free browser extension-based social web highlighter that you can use to capture content
online and highlight and organize quotes and ideas. In addition to its functions, the Glasp browser
extension provides transcripts of YouTube videos with one click with fewer timestamps.

In contrast to the YouTube default system that provides the transcript in short sentences and few
words per timestamp, the Glasp extension provides a concise and correct text format of the video
content with lesser timestamps by combining exact words spoken in at least 30 seconds of the
video. 20

20
https://blog.glasp.co/how-to-get-transcripts/
Pros and cons of Generative AI [Onaiza Maryam(23-SE-87)]
Benefits of Generative AI
Generative AI systems have many benefits. Let’s look at some of the best enhancements
that applications of generative AI enable.

1. Automates and innovates content generation:

One of the primary use cases many businesses see with generative AI is with content creation.
Marketing teams spend much of their time creating new content in the form of marketing copy,
blog posts, social media posts, and graphic design.

AI can help with all of this. Generative AI tools can take instructions for specific use cases. For
instance, if you want to create a landing page, tell your AI text generator to create an introduction
paragraph to your page identifying your customers’ pain points while tying to potential solutions
enabled by your product.
These features allow companies to not only automate content creation but also innovate. Give these
AI tools a shot by providing them with new ideas as input. See how they can take your suggestions
and come up with even more. You can then have a back-and-forth with the AI-generated ideas to
make them better, eventually leading to a usable idea.

2. Enhances customer experience through personalization:

Another way AI is improving business operations is by personalizing customer interactions. AI
can learn about your company and what you offer. And when you combine that with the customer
data you collect, you can create personalized experiences using generative AI.

For instance, an e-commerce business will have data on customer demographics and the products
they are likely to buy. If they can match those preferences to potential customers, those e-
commerce companies can create personalized content using AI with more relevant
recommendations.

Ultimately, this leads to a better customer experience because the customer gets the products most
suited to them.

3. Optimizes product designs:

Product design is another area where generative AI can help businesses improve productivity and
creativity. It’s not always easy to determine what customers want—behavior and preferences can
shift over time and lead businesses scrambling to keep up.

AI helps by analyzing large amounts of data for you. AI models use deep learning techniques to
identify market trends and analyze other market factors—helping companies gain more confidence
in their decisions and reduce risk. That data will help surface insights into consumer behavior that
your business can use to build new products and improve your current offering.

Once you understand where customers' preferences are shifting, use AI to develop ideas. Input
some of the new issues consumers face and possible solutions—as well as potential changes to
your current products that make them more appealing to the new market climate.

4. Strengthens cybersecurity efforts:

Generative AI has a role in helping businesses enhance their cybersecurity efforts. Companies must
analyze large amounts of data to look for threats—something AI tools can help with.
Although it’s possible for humans to analyze the information flowing in and out of a computer
network, this process takes a lot of time to do properly—time IT professionals can spend on other
tasks. AI helps by looking at that data for you and finding patterns of activity that fall outside the
norm. If something doesn’t look right, AI can inform your team to let them know of a possible
threat to address.

This approach leads to rapid detection of threats and the ability to stop bad actors before they
breach your internal systems. As cyberattacks begin to make more use of AI, this approach will be
vital for keeping up with more sophisticated threats that use generative AI to create new malware
and personalized phishing attacks.

5. Advances health care research:

Generative AI is helping advance health care research by allowing researchers to devise novel drug
ideas that would otherwise take extensive time and effort to do manually. For instance,
understanding the protein structure of molecules takes a long time—often years. But with
Google’s AlphaFold, this process happens within a few seconds.

This ability allows researchers to better understand molecules and how they interact —helping
them generate new drug compounds and test their effectiveness.

Another area generative AI helps in this process is by generating test data. In the past, researchers
needed human subjects to generate data for tests. But now that AI can generate synthetic patient
data for researchers, they can create tests in advance to determine if a drug has a chance before
starting human trials.

6. Streamlines business processes:

AI helps streamline business processes. If you can identify areas where you can automate tasks
and use AI to generate data, you can reduce the workload of your employees and get more done
each day.

An example of this is report analysis. Business managers must analyze reports detailing
information about their company and market. In order to digest all of this information, they spend
a lot of time analyzing the reports to get a complete understanding.
One of the great things about today’s large language models (LLMs) is their ability to analyze data
and draw conclusions. Many tools allow you to insert text reports into the AI text generator and
generate summaries of the text. Other software allows you to chat with PDFs to streamline the
summarization process.

7. Improved customer service:

Maintaining fast response times for customer issues can prove challenging when you have a lot of
customers. You only have so many team members, so customers may wait too long to speak to a
representative.

Now that chatbots use generative AI technology based on business data, you can offer personalized
support using AI tools. As a result, these tools can learn from your product information and
customer details to provide personalized support to people who need help.

Customers can talk to your AI chatbot 24/7 and get the help they need. If the chatbot can’t resolve
the issue, the customer is put in touch with a human representative—helping reduce the work your
team must do.

8. Fosters market innovation:

Companies looking to make a name in their industry can use generative AI to help the process. By
analyzing large amounts of data, generative AI tools can offer insights into data that humans may
not have the time to find on their own.

For businesses, this means more market innovation. AI helps companies find new avenues for
growth. These avenues include new product development, service opportunities, potential market
changes, and additional valuable insights.

On top of helping companies gain insights about markets, generative AI can also help reduce the
risks associated with innovation. Without understanding the data you have, you may not have all
the information required for the best decision-making.

The information you get from AI analysis will reduce your risk when developing new products
because you’ll understand more about consumer preferences. You’ll have a better indication of
how your idea will resonate with your target audience—giving your business a competitive
advantage.
9. Drives digital transformation:
Generative AI can help drive digital transformation in business because of the amount of data it
offers to companies, which helps leaders make better decisions.

For instance, a construction company may not have much interest in investing in technology. They
spend much of their time in the field, so they aren’t often using technology.

But when they start using machine learning AI algorithms to analyze their equipment and tell them
when something may go wrong, that changes. AI that offers predictive maintenance allows
companies to stay ahead of the curve and take care of equipment repairs before something goes
wrong—giving these companies a reason to invest in digital transformation.

10. Inspires creativity:

It’s not easy to come up with new ideas. Countless products and works of art already exist that
offer everything people need to survive and thrive in the world.

But that doesn’t mean there aren’t new ideas to explore—and that’s something generative AI can
help with.

Generative AI can help its users flesh out new ideas. Take AI chatbots, for example. Users of AI
chatbots can converse with these tools using natural language to get ideas that they can use in their
creative pursuits. For instance, a product designer can give a list of problems to a chatbot to have
it create a list of potential products that solve those issues.

You can also do this with art—give an AI art generator an idea to explore and have it create a
unique image. It may not be the idea you’re going for but it can give you a starting point to explore
new ideas.

Disadvantages of Generative AI
In addition to many of the known limitations outlined below, generative AI may be prone to
problems yet to be discovered or not fully understood

1. Large language models (LLMs) are prone to "hallucinations" - generating fictitious

information, presented as factual or accurate. This can include citations, publications,
biographical information, and other information commonly used in research and academic
papers.
 2. In addition to fictitious information, answers generated by LLMs can be wrong, often times
presented as correct or authoritative.

3. Due to their fundamental structure, as well as regularly released newer versions, content
from many generative AI models can be difficult to reproduce consistently. This is
particularly problematic in research and academia where reproducibility is a cornerstone of
establishing credibility.

4. Generative AI models are not databases of knowledge but rather an attempt to synthesize
and reproduce the information they have been trained on. This makes it incredibly difficult
to validate and properly attribute the basis of their content.

5. Many common generative AI tools are not connected to the internet and cannot update
or verify the content they generate.

6. The nature of generative AI models, particularly when given simple prompts, can be very
reductive, resulting in content that is over-simplified, low quality, or very generic.

7. Many generative AI models (including ChatGPT) are trained on data with cutoff dates,
resulting in outdated information or the inability to provide answers about current
information and events. In some cases, the data cutoff date is not made explicitly clear to
the user.

Future of Generative AI [Ifrah Sarfraz(23-SE-09)]

The creation of ever more potent algorithms with the ability to produce outputs that are more
nuanced is one of the most intriguing potential directions for generative AI in the future. This might
 encompass anything from virtual reality settings to music and art, and it has the power to alter how
we perceive and engage with technology.

Generative AI is already demonstrating its potential to revolutionize industries, with models like
Stable Diffusion and ChatGPT setting historical records for user growth. The market is witnessing
an influx of startups developing foundation models, AI-native apps, and infrastructure.

“Generative AI has the potential to change the world in ways that we can’t even imagine. It has
the power to create new ideas, products, and service that will make our lives easier, more
productive, and more creative. It also has the potential to solve some of the world’s biggest
problems, such as climate change, poverty, and disease. The future of generative AI is bright, and
I’m excited to see what it will bring.”21

However, the question remains: Where in this market will value accrue? The companies creating the most
value—i.e., training generative AI models and applying them in new apps—haven't captured most of it.
The future will likely see a shift in this dynamic, with value accruing in areas of the stack that are
differentiated and defensible.

21
Bill Gates, the co-founder of Microsoft, envision a bright future with generative AI as its core.
The generative AI market is expected to see a rise in infrastructure vendors, capturing the majority
of dollars flowing through the stack. As generative AI continues to evolve, we can anticipate a
surge in the development of AI applications, with a focus on improving retention, product
differentiation, and gross margins.

Massive investments are pouring into generative Al across multiple dimensions of human
endeavor. Venture capitalists, established corporations, and virtually every business in between are
investing generative Al startups at breakneck speed. The universal "magic" of LLMs is an uncanny
ability to mediate human interaction with big data, to help people make sense of information by
explaining it simply, clearly, and astonishingly fast. This suggests that generative Al will become
embedded in a multitude of existing applications and cause the invention of a second wave of new
applications.

Gartner, for example, predicts that 40% of enterprise applications will have embedded
conversational Al by 2024, 30% of enterprises will have Al-augmented development and testing
strategies by 2025, and more than 100 million workers will collaborate with "robocolleagues" by
2026.22

Of course, it's possible that the risks an limitations of generative Al will derail this steamroller.
Fine-tuning generative models to learn the nuances of what makes a business unique may prove
too difficult, running such computationally intensive models may prove too costly, and an
inadvertent exposure of trade secrets may scare companies away.

Or it all may happen but at a slower pace than many now expect. As a reminder, the promise of the
internet was realized, eventually. But it took a decade longer than the first generation of enthusiasts
anticipated, during which time necessary infrastructure was built or invented and people adapted
their behavior to the new medium's possibilities. In many ways, generative Al is another new
medium.

Influencers are thinking broadly about the future of generative Al in business.

"It may mean we build companies differently in the future," says Sean Ammirati, a venture
capitalist who is also the distinguished service professor of entrepreneurship at Carnegie Mellon

22
https://www.oracle.com
University's Tepper School of Business and cofounder of CMU's Corporate Startup Lab. In the
same way that "digital native” companies had an advantage after the rise of the internet, Ammirati
envisions future companies built from the ground up on generative Al-powered automation will be
able to take the lead.

"These companies will be automation-first, so they won't have to relearn how to stop doing things
manually that they should be doing in an automated way," he says. "You could end up with a very
different kind of company."

The increasing emphasis on moral and responsible AI development is another significant

development to monitor in the future of generative AI. We must approach AI development and use
it responsibly, since it has the potential to have a significant impact on society. This involves
making sure that AI is developed with transparency and comprehensibility in mind, and that it is
deployed in ways that benefit society.

The future of generative AI also holds potential pitfalls. While impressive, the rapid growth of
generative AI applications is not enough to build durable software companies. The challenge lies
in ensuring that growth is profitable and that users generate profits once they sign up and stick
around for a long time.
Generative AI will undoubtedly have a significant impact on how society and technology develop
in the future. In the years to come, we can anticipate seeing even more intriguing and cutting-edge
applications emerge as more researchers and developers continue to explore this area.

In conclusion, the future of generative AI is promising, but it's not without its challenges. As we
continue to explore its potential, we must also address the issues of value accrual, profitability, and
retention. The journey ahead is exciting, and the impact of generative AI on our lives and the
market is bound to be profound.

CONCLUSION:

Generative Al has the potential to transform our world in profound ways. As we work towards
harnessing its full potential, it is essential that we remain vigilant and responsible in our approach.

Through continued innovation, research, and collaboration, we can unlock the full power of
Generative Al and drive unprecedented progress.

In this contemplative conclusion, we pause to reflect on Al's remarkable inflection point and the
unprecedented rise of Generative Al. The journey of Al from a theoretical concept to a tangible
force that shapes our daily lives has been nothing short of extraordinary. The advent of Al
Generative marks a turning point, where machines evolve from passive tools to creative
collaborators.

Generative Al's ascendancy has unleashed a wave of innovation, pushing the boundaries of human-
machine interactions. As we look back on Al's transformative journey, it becomes evident that
Generative Al embodies the convergence of Al's evolution and human aspirations. Its rise signifies
the dawn of a new era, where technology not only aids us but also inspires us to explore uncharted
territories of creativity and imagination.

In this forward-looking perspective, we recognize the importance of nurturing curiosity and

fostering innovation in the ever-expanding Al landscape. As Al Generative continues to evolve, a
spirit of curiosity becomes a driving force for exploration and learning. By encouraging continuous
research and experimentation, we unlock Al's latent potential and pave the way for groundbreaking
discoveries.
Nurturing innovation requires a collaborative mindset, where diverse perspectives converge to
address complex challenges and propel Al's growth. By fostering a culture that values creativity,
risk- taking, and interdisciplinary collaboration, we lay the foundation for a future where Al-driven
solutions transcend barriers and enrich human experiences.

In conclusion, the rise of Generative Al signifies an exciting inflection point in the journey of Al's
evolution. Embracing this transformative era calls for action, curiosity, and a commitment to
ethical practices. By embracing Generative Al's limitless potential, we can unlock new
possibilities, redefine user experiences, and create a more inclusive and prosperous future for all.
Let us unite in our pursuit of excellence and shape a future where technology, innovation, and
human aspirations intertwine to create a world of boundless possibilities.