T32 AI Trends, Technologies, & Transformations 250323.pdf

Artificial Intelligence: Trends,
Technologies, and Transformations
1
Moustafa M. Elsayed
EGEC Engineering House of Expertise
Giza, Egypt
March 2025

Preface
2
Artificial Intelligence (AI) has transitioned from a futuristic concept to a
transformative force that permeates nearly every aspect of our daily lives.
From healthcare and finance to manufacturing and transportation. AI is
reshaping industries, driving innovation, and redefining the boundaries of
what technology can achieve. The 21st century has witnessed remarkable
advancements in AI, particularly in machine learning and deep learning,
fueled by the availability of vast datasets and powerful computing
resources. These advancements have enabled AI systems to recognize
patterns, process natural language, and make informed decisions with
unprecedented accuracy.
This presentation, aims to provide a comprehensive overview of AI's
current state, its historical evolution, and its future trajectory. We will
explore the fundamental components of AI, delve into the intricacies of
machine learning and neural networks, and examine the diverse
applications of AI across various industries. Additionally, we will discuss the
tools and technologies driving AI innovation, as well as the ethical, social,
and environmental concerns that accompany its rapid growth.

Preface
3
As we stand on the brink of a new era defined by intelligent systems, it is
crucial to understand both the opportunities and challenges that AI
presents. This presentation seeks to equip you with the knowledge to
navigate the complexities of AI, appreciate its potential, and critically
assess its impact on society.
Welcome to the world of Artificial Intelligence.
Moustafa M. Elsayed
EGEC Engineering House of Expertise
Giza, Egypt March 2025
Note: This preface was prepared by https://chat.deepseek.com/ for my
presentation.

‫تقديم‬
4
‫انتقل‬
‫جا‬ ‫كل‬ ‫ا‬ً‫ب‬‫تقري‬ ‫تغطي‬ ‫تحويلية‬ ‫قوة‬ ‫إلى‬ ‫مستقبلي‬ ‫مفهوم‬ ‫مجرد‬ ‫كونه‬ ‫من‬ ‫االصطناعي‬ ‫الذكاء‬
‫جوانب‬ ‫من‬ ‫نب‬
‫اليومية‬ ‫حياتنا‬
،
‫والنقل‬ ‫التصنيع‬ ‫إلى‬ ‫والتمويل‬ ‫الصحية‬ ‫الرعاية‬ ‫من‬
.
‫تش‬ ‫االصطناعي‬ ‫الذكاء‬ ‫يعيد‬
،‫الصناعات‬ ‫كيل‬
‫التكنولوجيا‬ ‫تحققه‬ ‫أن‬ ‫يمكن‬ ‫ما‬ ‫حدود‬ ‫تعريف‬ ‫ويعيد‬ ،‫االبتكار‬ ‫ويقود‬
.
‫و‬
‫والع‬ ‫الحادي‬ ‫القرن‬ ‫شهد‬ ‫لقد‬
‫ا‬ً‫م‬‫تقد‬ ‫شرون‬
‫بتو‬ ‫ا‬ً‫م‬‫مدعو‬ ،‫العميق‬ ‫والتعلم‬ ‫اآللي‬ ‫التعلم‬ ‫مجاالت‬ ‫في‬ ‫خاصة‬ ،‫االصطناعي‬ ‫الذكاء‬ ‫مجال‬ ‫في‬ ‫ا‬ً‫ظ‬‫ملحو‬
‫فر‬
‫قوية‬ ‫حوسبة‬ ‫وموارد‬ ‫ضخمة‬ ‫بيانات‬ ‫مجموعات‬
.
‫من‬ ‫االصطناعي‬ ‫الذكاء‬ ‫أنظمة‬ ‫التطورات‬ ‫هذه‬ ‫مكنت‬ ‫وقد‬
‫ومعالجة‬ ،‫األنماط‬ ‫على‬ ‫التعرف‬
‫الطبيعية‬ ‫اللغات‬
‫مسبوقة‬ ‫غير‬ ‫بدقة‬ ‫مستنيرة‬ ‫قرارات‬ ‫واتخاذ‬ ،
.
‫ت‬
‫هدف‬
‫المحاضرة‬ ‫هذه‬
‫التاريخي‬ ‫وتطوره‬ ،‫االصطناعي‬ ‫للذكاء‬ ‫الحالي‬ ‫الوضع‬ ‫عن‬ ‫شاملة‬ ‫نظرة‬ ‫تقديم‬ ‫إلى‬
،
‫المستقبلي‬ ‫ومساره‬
.
‫المحاضرة‬ ‫تكشف‬
‫و‬ ،‫االصطناعي‬ ‫للذكاء‬ ‫األساسية‬ ‫المكونات‬
‫ت‬
‫تفاص‬ ‫في‬ ‫خوض‬
‫اآللي‬ ‫التعلم‬ ‫يل‬
‫الصناعات‬ ‫مختلف‬ ‫عبر‬ ‫االصطناعي‬ ‫للذكاء‬ ‫المتنوعة‬ ‫التطبيقات‬ ‫وفحص‬ ،‫العصبية‬ ‫والشبكات‬
.
‫ب‬
‫إلى‬ ‫اإلضافة‬
،‫ذلك‬
‫ت‬
‫ناقش‬
‫المحاضرة‬
‫وكذلك‬ ،‫االصطناعي‬ ‫الذكاء‬ ‫مجال‬ ‫في‬ ‫االبتكار‬ ‫تقود‬ ‫التي‬ ‫والتقنيات‬ ‫األدوات‬
‫المخاوف‬
‫السريع‬ ‫نموه‬ ‫ترافق‬ ‫التي‬ ‫والبيئية‬ ‫واالجتماعية‬ ‫األخالقية‬
.
‫و‬
‫جديد‬ ‫عصر‬ ‫أعتاب‬ ‫على‬ ‫نقف‬ ‫بينما‬
‫ل‬
‫الذكية‬ ‫ألنظمة‬
‫من‬ ‫فإنه‬
‫الت‬ ‫والتحديات‬ ‫الفرص‬ ‫فهم‬ ‫الضروري‬
‫يقدمها‬ ‫ي‬
‫االصطناعي‬ ‫الذكاء‬
.
‫ت‬ ‫و‬
‫سعى‬
‫المحاضرة‬
‫ا‬ ‫الذكاء‬ ‫تعقيدات‬ ‫مع‬ ‫للتعامل‬ ‫الالزمة‬ ‫بالمعرفة‬ ‫تزويدك‬ ‫إلى‬
،‫الصطناعي‬
‫نقدي‬ ‫بشكل‬ ‫المجتمع‬ ‫على‬ ‫تأثيره‬ ‫وتقييم‬ ،‫إمكاناته‬ ‫وتقدير‬
.
‫االصطناعي‬ ‫الذكاء‬ ‫عالم‬ ‫في‬ ‫بكم‬ ‫ا‬ً‫ب‬‫مرح‬
.
‫م‬ ‫مصطفى‬
.
‫السيد‬
،
‫الهندسي‬ ‫الخبرة‬ ‫بيت‬
EGEC
،
‫مصر‬ ،‫الجيزة‬
،
‫مارس‬
2025
‫مالحظة‬
:
‫التقديم‬ ‫هذا‬ ‫إعداد‬ ‫تم‬
‫اإلنجليزية‬ ‫باللغة‬ ‫أوليا‬
‫بواسطة‬
https://chat.deepseek.com/
‫ل‬
،‫لمحاضرة‬
‫جانبي‬ ‫من‬ ‫طفيفة‬ ‫تعديالت‬ ‫بعد‬ ‫هنا‬ ‫وقدم‬ ،‫الموقع‬ ‫بنفس‬ ‫العربية‬ ‫اللغة‬ ‫إلى‬ ‫الترجمة‬ ‫تم‬ ‫ثم‬
.

Contents
• Preface
• Statistics for AI Usage in Y2025
• Introduction to AI
• Basic Components of AI
• Main Branches of AI
• History of AI
• Machine Learning
• Machine Learning vs Neural
Networks vs Deep Learning
• Machine Learning Models
• Neural Networks Basics
• Concept
• Definitions
• Neural Network Operation
• Learning Process
5
• Types of Neural Networks
• Applications of Neural Networks
• Advantages & Disadvantages
• AI Tools
• What is AI Tool?
• Generative AI Applications
• Examples of AI Tools
• Concerns of Growth of AI
Technology
• References

Statistics for AI Usage in Y2025
6
• Artificial intelligence is reshaping the world at an unprecedented
rate, accelerating progress like never before.
• See comprehensive overview statistical report by
https://juliety.com/ai-statistics for year 2025.
• The report is based on fresh reports and the latest research from
industry leaders and has everything needed to understand the
trends of AI in 2025, its profound implications, and how it’s
redefining the boundaries of technology and innovation.
• The results shown in the following pages for statistics serve as a
foundation for understanding the dynamic, disruptive power of AI
now and in the years to come.
https://juliety.com/ai-statistics
About the Statistics of AI Usage in Y2025

7

8

9

10

Review Questions (Prepared by Deepseek)
11
1. What are the most common use cases for AI in 2025?
2. What percentage of top-level US business leaders are "often using AI"?
3. What are the top concerns related to AI adoption?
4. How do companies plan to invest in AI over the next year?
5. What are the three main attitudes toward AI in the workplace?
6. What is the predicted impact of AI on the job market?
7. Which AI use case has the highest adoption rate among businesses in 2025?
8. What are the biggest concerns about AI, such as technology dependence and privacy?
9. How is AI being used in customer support, and what are the most common
applications?
10. What percentage of businesses are "interested but not using AI yet"?

Introduction to AI
About Artificial Intelligence (AI)
12
• At its core, AI is the broad concept
of enabling machines to perform
tasks that typically require human
intelligence.
• This encompasses abilities like
reasoning, learning, problem-
solving, perception, and language
understanding.
• AI is used in numerous
applications, from virtual assistants
and chatbots to self-driving cars
and medical diagnosis.

Introduction to AI
13
• Machine learning, involves creating models by training an algorithm
to make predictions or decisions based on data.
• It encompasses a broad range of techniques that enable computers
to learn from and make inferences based on data without being
explicitly programmed for specific tasks.
• One of the most popular types of machine learning algorithm is
called a neural network (or artificial neural network).
• Neural networks are modeled after the human brain's structure and
function. A neural network consists of interconnected layers of
nodes (analogous to neurons) that work together to process and
analyze complex data.
• Neural networks are well suited to tasks that involve identifying
complex patterns and relationships in large amounts of data.
https://www.ibm.com/think/topics/artificial-intelligence
Machine Learning

Introduction to AI
14
• Learning
• Reasoning and Decision
Making
Basic Components of AI
• Problem Solving
• Perception
• Language Understanding

Introduction to AI
15
• Learning is a crucial component of AI as it enables AI systems to
learn from data and improve performance without being explicitly
programmed by a human.
• AI technology learns by labeling data, discovering patterns within
the data, and reinforcing this learning via feedback.
• Examples include voice recognition systems like Siri or Alexa, learn
correct grammar and the skeleton of a language, identify objects.
• One fundamental aspect of AI learning is the trial-and-error
method where AI system attempts various solutions to a problem
and retains successful strategies in its database for future use.
https://www.soci.ai/knowledge-articles/branches-of-artificial-intelligence/
https://ellow.io/components-of-ai/
Basic Components of AI: Learning

Introduction to AI
16
• AI systems can use logical rules, probabilistic models, and
algorithms to draw conclusions and make inferred decisions.
• When faced with problems or issues, AI models should use
reasoning to generate consistent results.
• Example include writing assistant to improve language style.
• Example is chess-playing programs use reasoning to evaluate
possible moves and make decisions based on the likely
outcomes.
Basic Components of AI: Reasoning and Decision Making

Introduction to AI
17
• AI systems take in data, manipulate it and apply it to create a
solution that solves a specific problem.
• Example: Route optimization algorithms in navigation systems
solve the problem of finding the most efficient path between two
points.
• AI’s problem-solving ability involves techniques like planning,
search, and optimization.
Basic Components of AI: Problem Solving

Introduction to AI
18
• Perception is the ability to see, hear, or become aware of
something through the senses.
• AI perception is crucial for tasks like image recognition, object
detection, image segmentation, and video analysis.
• Example: Self-driving cars gather visual data to recognize roads,
lanes, and obstacles and then map these objects.
Basic Components of AI: Perception

Introduction to AI
19
• Natural language understanding (NLU) enables human-computer
interaction.
• NLU enables computers to understand natural language used by
humans, such as English, French or Arabic.
• One of the main purposes of NLU is to create chat- and voice-
enabled bots that can interact with people without supervision.
• A basic form of NLU is called parsing, which takes written text
and converts it into a structured format for computers to
understand.
https://www.techtarget.com/searchenterpriseai/definition/natural-language-understanding-NLU
Basic Components of AI: Language Understanding

Introduction to AI
20
• Machine Learning (ML)
• Deep Learning (DL)
• Natural Language Processing (NLP)
Main Branches of AI
????
• Robotics
• Fuzzy Logic

Introduction to AI
21
• ML is the ability of machines to learn from data and algorithms
automatically.
• The ML algorithm could be divided into three main parts.
• A Decision Process: ML algorithms are used to make a prediction or
classification. Based on some input data, which can be labeled or
unlabeled, the algorithm will produce an estimate about a pattern in the
data.
• An Error Function: An error function evaluates the prediction of the
model. If there are known examples, an error function can make a
comparison to assess the accuracy of the model.
• A Model Optimization Process: If the model can fit better to the data
points in the training set, then weights are adjusted to reduce the
discrepancy between the known example and the model estimate. The
algorithm will repeat this iterative “evaluate and optimize” process,
updating weights autonomously until a threshold of accuracy has been
met.
Main Branches of AI: Machine Learning (ML)
https://www.ibm.com/think/topics/machine-learning

Introduction to AI
22
• DL is a subset of machine learning.
• DL uses neural networks to extract abstract features from the
data, leading to better performance than machine learning and
often more powerful representations.
• DL needs a much greater volume of data than ML.
• Examples are home assistance technology like Amazon Alexa.
Main Branches of AI: Deep Learning (DL)

Introduction to AI
23
• NLP allows computers to understand spoken words and written text.
• NLP is the most commonly used AI as it’s intertwined in many of
today’s digital assistants, chatbots, virtual assistants, and spam
detection.
• NLP is also used to generate sentiment analysis, which analyzes texts
and extracts the emotions and attitudes about a product or service.
Main Branches of AI: Natural Language Processing (NLP)

Introduction to AI
24
• Robotics utilizes AI to develop and design robots or machines
capable of performing tasks autonomously or semi-autonomously.
• Robotics involves other components of AI technology, such as NLP,
ML, or perception.
• AI-based robots are already in many industries, such as healthcare,
retail, and manufacturing.
Main Branches of AI: Robotics

Introduction to AI
25
• Fuzzy logic is an approach to computing based on "degrees of
truth" rather than the usual "true or false" (1 or 0).
• The term fuzzy refers to things that are not clear or are vague. In
the real world many times we encounter a situation when we can’t
determine whether the state is true or false.
• For Example, fuzzy logic can help your automatic braking system
determine how hard it should brake.
• Fuzzy Logic is a form of many-valued logic in which the truth values
of variables may be any real number between 0 and 1, instead of
just the traditional values of true or false.
Main Branches of AI: Fuzzy Logic
https://www.geeksforgeeks.org/fuzzy-logic-introduction/

Introduction to AI
26
More branches are continuously being added to AI technology,
reflecting the progress we witness day by day in the field of AI.
https://www.geeksforgeeks.org/top-10-branches-of-artificial-intelligence/
Main Branches of AI: Other Branches

Introduction to AI
27
1. What are the basic components of AI, and how do they contribute to AI systems?
2. How does machine learning differ from deep learning, and what are their key
characteristics?
3. What is the role of neural networks in AI, and how are they modeled after the human
brain?
4. How do AI systems use reasoning and decision-making to solve problems?
5. What are some examples of AI applications in perception, such as image recognition or
self-driving cars?
6. How does natural language understanding (NLU) enable human-computer interaction?
7. What are the main branches of AI, and how do they differ from one another?
8. How does fuzzy logic differ from traditional binary logic, and what are its real-world
applications?
9. What is the significance of robotics in AI, and how is it applied in industries like
healthcare and manufacturing?
10. How do AI systems learn from data, and what methods do they use to improve
performance over time?

History of AI
28
For further details on information given in this section see
https://www.tableau.com/data-insights/ai/history
• In ancient times, inventors made things called “automatons”
which were mechanical and moved independently of human
intervention.
• The word “automaton” comes from ancient Greek, and means
“acting of one’s own will.”
• One of the earliest records of an automaton comes from 400 BCE
and refers to a mechanical pigeon.
• One of the most famous automatons was created by Leonardo da
Vinci around the year 1495.
• Only in the 20th century, engineers and scientists began to make
strides toward our modern-day AI.
Before The 20th Century

History of AI
29
• 1921: a science fiction play “Rossum’s Universal Robots” which
introduced the idea of “artificial people” which he named robots.
This was the first known use of the word.
• 1929: Japan, the first Japanese robot, named Gakutensoku.
• 1949: USA, Published the book “Giant Brains, or Machines that
Think” which compared the newer models of computers to human
brains.
Years 1900 to 1950

History of AI
30
• 1950: Publishing “Computer Machinery and Intelligence” which
proposed a test of machine intelligence called The Imitation
Game.
• 1952: Developing a program to play checkers, which is the first to
ever learn the game independently.
• 1955: A workshop on “artificial intelligence” which is the first use
of the word, and how it came into popular usage.
Birth of AI: 1950-1956

History of AI
31
• 1958: the first programming language for AI research.
• 1959: Using the term “machine learning” for first time.
• 1961: The first industrial robot Unimate in assembly line at General
Motors in New Jersey.
• 1965: The first “expert system” first trial to replicate the thinking
and decision-making abilities of human experts.
• 1966: The first “chatterbot” (later shortened to chatbot), ELIZA,
using natural language processing (NLP) to converse with humans.
• 1968: Publishing “Group Method of Data Handling” proposing a
new approach to AI that would later become as “Deep Learning.”
• 1979: The American Association of Artificial Intelligence which is
now known as the Association for the Advancement of Artificial
Intelligence (AAAI) was founded.
AI Maturation: 1957-1979

History of AI
32
Learning techniques and the use of Expert System became more
popular, allowing computers to learn from their mistakes.
• 1980: The first expert system in commercial market, known as
XCON.
• 1981: The Japanese government allocated $850 million to create
computers that could translate, converse in human language, and
express reasoning on a human level.
• 1986: Munich, first driverless car (or robot car). It could drive up to
55 mph on roads that didn’t have other obstacles or human drivers.
AI Boom: 1980-1987
AI winter: 1987-1993
Both private investors and the government lost interest in AI and
halted their funding due to high cost versus seemingly low return.

History of AI
33
• 1997: Deep Blue by IBM beats the world chess champion.
• 1997: Windows released a speech recognition software.
• 2000: robot that could simulate human emotions with its face,
which included eyes, eyebrows, ears, and a mouth. It was called
Kismet.
• 2002: first Roomba (vacuum cleaner robot).
• 2010: first gaming hardware to track body movement and translate
it into gaming directions.
• 2011: An NLP computer programmed to answer questions
named Watson (created by IBM) won Jeopardy against two former
champions in a televised game.
• 2011: Apple released Siri, the first popular virtual assistant.
AI Agents: 1993-2011

History of AI
34
• 2012: Neural network to recognize cats
• 2016: Sophia, first robot with a realistic human appearance to see
and replicate emotions, as well as to communicate.
• 2017: AI chatbots to converse and to negotiate.
• 2018: A Chinese AI beats human intellect on a Stanford reading
and comprehension test.
• 2019: Google’s AlphaStar reached Grandmaster on the video game
StarCraft 2, outperforming all but .2% of human players.
• 2020: GPT-3 to create code, poetry, and other such language and
writing tasks.
• 2021: DALL-E, to process and understand images enough to
produce accurate captions, moving AI one step closer to
understanding the visual world.
AI 2012-present

Machine Learning
35
• Neural networks is a sub-
field of machine learning,
and deep learning is a
sub-field of neural
networks.
• "Deep" machine learning
can use labeled datasets,
also known as supervised
learning, to inform its
algorithm, but it doesn’t
necessarily require a
labeled dataset.
https://www.ibm.com/think/topics/machine-learnng
Machine Learning vs Neural Networks vs Deep Learning

Machine Learning
36
• Classical, or "non-deep," machine learning is more dependent on
human intervention to learn.
• Neural networks, or artificial neural networks (ANNs), are
comprised of node layers, containing an input layer, one or more
hidden layers, and an output layer.
• A neural network that consists of more than three layers—which
would be inclusive of the input and the output—can be considered
a deep learning algorithm or a deep neural network.
• A neural network that only has three layers is just a basic neural
network.
• Deep learning and neural networks are credited with accelerating
progress in areas such as computer vision, natural language
processing (NLP), and speech recognition.
Machine Learning vs Neural Networks vs Deep Learning

Machine Learning
37
Supervised Learning
• Supervised learning use of labeled datasets to train algorithms to
classify data or predict outcomes accurately.
• As input data is fed into the model, the model adjusts its weights
until it has been fitted appropriately.
Machine Learning Models
Unsupervised Learning
• Unsupervised learning uses machine learning algorithms to
analyze and cluster unlabeled datasets.
• These algorithms discover hidden patterns or data groupings
without the need for human intervention.
• Unsupervised learning’s make it ideal for exploratory data
analysis, cross-selling strategies, customer segmentation, and
image and pattern recognition.

Machine Learning
38
Semi-Supervised Learning
• During training, it uses a smaller labeled data set to guide
classification and feature extraction from a larger unlabeled data set.
• Semi-supervised learning can solve the problem of not having
enough labeled data for a supervised learning algorithm. It also
helps if it’s too costly to label enough data.
Reinforcement Learning
• This is a model that is similar to supervised learning, but the
algorithm isn’t trained using sample data.
• This model learns as it goes by using trial and error.
• A sequence of successful outcomes will be reinforced to develop the
best recommendation or policy for a given problem.
Machine Learning Models

Machine Learning
39
Common Machine Learning Algorithms
Neural Networks
• Neural networks simulate the way the human brain works, with a
huge number of linked processing nodes.
Linear regression
• This algorithm is used to predict numerical values, based on a linear
relationship between different values.
• For example, the technique could be used to predict house prices
based on historical data for the area.

Machine Learning
40
Logistic Regression
• This supervised learning algorithm makes predictions for categorical
response variables, such as “yes/no” answers to questions.
• It can be used for applications such as classifying spam and quality
control on a production line.
Clustering
• Using unsupervised learning, clustering algorithms can identify
patterns in data so that it can be grouped.
• Computers can help data scientists by identifying differences
between data items that humans have overlooked.

Machine Learning
41
Decision trees
• Decision trees can be used for both predicting numerical values
(regression) and classifying data into categories.
• Decision trees use a branching sequence of linked decisions that can
be represented with a tree diagram.
• One of the advantages of decision trees is that they are easy to
validate and audit, unlike the black box of the neural network.
Random forests
• In a random forest, the machine learning algorithm predicts a value
or category by combining the results from a number of decision
trees.

Machine Learning
42
1. What is the relationship between machine learning, neural networks, and deep learning?
2. How does supervised learning differ from unsupervised learning in machine learning?
3. What are the key characteristics of a neural network, and how do they mimic the human
brain?
4. What is the purpose of semi-supervised learning, and how does it address the challenge
of labeled data?
5. How does reinforcement learning work, and what makes it different from supervised
learning?
6. What are some common machine learning algorithms, and how are they used in real-
world applications?
7. How does a decision tree algorithm work, and what are its advantages over other models
like neural networks?
8. What is the role of clustering in unsupervised learning, and how does it help identify
patterns in data?
9. How do random forests combine multiple decision trees to improve prediction accuracy?

Neural Networks Basics
43
• Neural networks are machine learning models that mimic the
complex functions of the human brain.
• These models consist of interconnected nodes or neurons that
process data, learn patterns, and enable tasks such as pattern
recognition and decision-making.
https://www.geeksforgeeks.org/neural-networks-a-beginners-guide/
Concept

44
• Neurons: The basic units that receive inputs. Each neuron has a
threshold and an activation function for its output.
• Connections: Links between neurons that carry information,
regulated by weights and biases.
• Weights and Biases: These parameters determine the strength
and influence of connections.
• Propagation Functions: Mechanisms that help process and
transfer data across layers of neurons.
Definitions- Key Components

45
• Input Layer: Each input neuron in the layer corresponds to a
feature in the input data.
• Hidden Layers: One or multiple hidden layers. Each layer consists
of units (neurons) that transform the inputs into something that
the output layer can use.
• Output Layer: The final layer produces the output of the model.
The format of these outputs varies depending on the specific task.
Definitions - Layers

46
https://gemini.google.com/app
Definitions - Neuron Threshold & Activation Function
• In biological neurons, a certain level of electrical potential must
be reached (the threshold) before the neuron "fires" and sends a
signal. This "all-or-nothing" principle is a fundamental aspect of
neural communication.
• The idea of a threshold in artificial neurons is an attempt to
replicate this behavior.
• Without activation functions, a neural network would simply be
a series of linear transformations.
• Real-world data is often highly non-linear.
• Activation functions introduce non-linearity, allowing neural
networks to learn complex patterns and relationships.

47
• A sigmoid function is any mathematical function whose graph has a
characteristic S-shaped or sigmoid curve.
• A common example of a sigmoid function is the logistic function.
There are many other mathematical functions.
https://medium.com/towards-data-science/first-neural-network-for-beginners-explained-with-code-4cfd37e06eaf
https://en.wikipedia.org/wiki/Sigmoid_function
Definitions - Sigmoid Function
Convolutional
Convolutional
Convolutional

48
• Neural networks can usually be read from left to right.
• The first layer is the layer in which inputs are entered.
• There are two or more internals layers (called hidden layers) that do
some math, and one last layer that contains all the possible outputs.
Neural Network Operation - 01

49
• The operation of a complete neural network is straightforward : one
enter variables as inputs (for example an image of a cat if the neural
network is supposed to tell what is on an image), and after some
calculations, an output is returned (giving an image of a cat should
return the word “cat”).
• So, we can represent
an artificial neural
network like that is
shown in the figure

50
• On the Figure, there are 3 inputs (x1, x2, x3) coming to the neuron.
• This value is multiplied, before being added, by another variable
called “weight” (w1, w2, w3) which determines the connection
between the two neurons.
• Each connection of neurons has its own weight, and those are the
only values that will be modified during the learning process.
• A bias value is added to the total value calculated. It is not a value
coming from a specific neuron and is chosen before the learning
phase. Here a value of 1 is used as a bias.

51
• At the end, the neuron applies a function called “activation
function” to the obtained value.
• The activation function serves to turn the total value calculated
before to a number between 0 and 1 (done for example by a
sigmoid function). Other functions exist.
• Conclusions: Take all values from connected neurons multiplied by
their respective weight, add them, add a bias and apply an
activation function. Then, the neuron is ready to send its new value
to other neurons.
• After every neurons of a column did it, the neural network passes to
the next column.
• In the end, the last values obtained should be one usable to
determine the desired output.

52
• Every inputs come with its label, explaining what output the neural
network should have guessed.
• If the obtained output doesn’t match the label, weights and biases
are changed.
• Those are the only variables that can be changed during the
learning phase.
• To determine which weight or bias is better to modify, a particular
process, called “backpropagation” is done. Inspect every connection
to check how the output would behave according to a change on
the weight.
• Modifying the weights and biases determines the speed the neural
network will learn, or more specifically how the network will modify
a weight, little by little or by bigger steps.
Learning Process – Weights & Biases

53
• Data passes from the input layer through the hidden layers to the
output layer.
• Network performance is measured as the difference between the
actual output and the predicted output, this is a loss function which
is a measure of error in the predictions. The loss function could
vary; common choices are mean squared error for regression.
• The network computes the gradients of the loss function with
respect to each weight and bias in the network to find out how
much each part of the output error can be attributed to each
weight and bias.
• The weights and biases are updated using an optimization
algorithm like stochastic gradient descent (SGD). The size of the
step taken in each update is determined by the learning rate.
Learning Process – Loss Function

54
• The process of forward propagation, loss calculation,
backpropagation, and weight update is repeated for many
iterations over the dataset.
• Over time, this iterative process reduces the loss, and the
network’s predictions become more accurate.
• Through these steps, neural networks can adapt their parameters
to better approximate the relationships in the data, thereby
improving their performance on tasks such as classification,
regression, or any other predictive modeling.
Learning Process - Iteration

55
• Learning in neural networks follows a three-stage process:
• Data is fed into the network.
• Based on the current parameters, the network generates an
output.
• The network refines its output by adjusting weights and
biases, gradually improving its performance.
• Training a neural network means calculating the loss (error value)
of the model and checking if it is reduced or not.
• If the error is higher than the expected value, update the model
parameters, such as weights and bias values.
• Use the model once the loss is lower than the expected error
margin.
Learning Process - Summary

56
• Feedforward Networks: Data moves from input to output in a
single direction.
• Singlelayer Perceptron: There is only one layer of neurons .
• Multilayer Perceptron (MLP): This is a feedforward neural network
with three or more layers, including input and output layers.
• Convolutional Neural Network (CNN): This is a specialized artificial
neural network designed for image processing. Convolutional
Neural Networks are widely used in image recognition and natural
language processing areas.
• Recurrent Neural Network (RNN): This is a neural network type
intended for sequential data processing. It is appropriate for
applications where contextual dependencies are critical.
Types of Neural Networks - 1

57
• The Convolutional Neural network (CNN) and Recurrent neural
network (RNN) can be considered as two of the most prominent
types of neural networks, which form the basis for most pre-
trained models in neural networks.
• CNN is a supervised learning model which consists of one or more
convolutional layers. First those convolutional layers apply a
convolutional function on the input. Then these layers are sent to
the next layer. The neurons in a layer do not necessarily need to
connect with the complete set of neurons in the next layer.
• RNN is a widely used neural network mostly used for speech
recognition and natural language processing (NLP). It recognizes
sequential characteristics of data and uses patterns to predict the
following scenario. In RNN, the output of the previous step
performs as the input to the next step.
Types of Neural Networks - 2

58
• Image and Video Recognition: CNNs are extensively used in
applications such as facial recognition, autonomous driving, and
medical image analysis.
• Natural Language Processing (NLP): RNNs and transformers power
language translation, chatbots, and sentiment analysis.
• Finance: Predicting stock prices, fraud detection, and risk
management.
• Healthcare: Neural networks assist in diagnosing diseases,
analyzing medical images, and personalizing treatment plans.
• Gaming and Autonomous Systems: Neural networks enable real-
time decision-making, enhancing user experience in video games
and enabling autonomous systems like self-driving cars.
Applications of Neural Networks

59
Advantages
• Neural networks are useful for activities where the link between
inputs and outputs is complex or not well defined because they can
adapt to new situations and learn from data.
• Their proficiency are in applications of pattern recognition in tasks
like audio and image identification, natural language processing,
and other intricate data patterns.
• Because neural networks are capable of parallel processing, they
can process numerous jobs at once, which speeds up and improves
the efficiency of computations.
Advantages & Disadvantages

60
Disadvantages
• Large neural network training demands a lot of computing power.
• As “black box” models, neural networks pose a problem in some
important applications since it is difficult to understand how they
make decisions. For example, when you put an image of a cat into a
neural network and it predicts it to be a car, it is very hard to
understand what caused it to arrive at this prediction.
• For efficient training, neural networks frequently need sizable,
labeled datasets; otherwise, their performance may suffer from
incomplete or skewed data.
Advantaged & Disadvantages

Review Questions
61
1. What are the key components of a neural network, and how do they function together?
2. How do activation functions introduce non-linearity into neural networks, and why is this
important?
3. What is the role of weights and biases in a neural network, and how are they adjusted during
the learning process?
4. How does the backpropagation algorithm work, and why is it essential for training neural
networks?
5. What is the purpose of a loss function in neural networks, and how does it guide the learning
process?
6. What are the differences between feedforward networks, convolutional neural networks
(CNNs), and recurrent neural networks (RNNs)?
7. How do convolutional neural networks (CNNs) process image data, and what makes them
suitable for tasks like image recognition?
8. What are the advantages of neural networks in tasks like pattern recognition and parallel
processing?
9. What are the main disadvantages of neural networks, particularly in terms of computational
requirements and interpretability?
10. How are neural networks applied in real-world scenarios such as healthcare, finance, and
natural language processing (NLP)?

AI Tools
62
• An AI tool is a software application or platform that utilizes
artificial intelligence (AI) technologies, such as machine learning,
natural language processing, computer vision, or deep learning, to
perform specific tasks or solve problems.
• These tools are designed to automate processes, analyze data,
make predictions, or enhance decision-making with minimal
human intervention.
• AI tools are used across various industries and applications.
What is AI Tool?

AI Tools
63
• An AI Artifact is a machine learning term that is used to describe the
output created by the training process.
• AI Artifacts describe all digital products that are used in an AI Tool.
• Generative AI can learn from existing artifacts to generate new.
• It can produce a variety of novel content, such as images, video,
music, speech, text, software code and product designs.
• Generative AI uses a number of techniques that continue to evolve.
https://www.gartner.com/en/topics/generative-ai#:~:text=for%20IT%20Leaders-
,What%20is%20generative%20AI%3F,software%20code%20and%20product%20designs.
Generative AI Applications

AI Tools
64
https://www.gartner.com/en/topics/generative-ai#:~:text=for%20IT%20Leaders-
,What%20is%20generative%20AI%3F,software%20code%20and%20product%20designs.

AI Tools
65
• Chatbots and Virtual Assistants: Tools like ChatGPT, DeepSeek,
Google Assistant, or Siri that use natural language processing (NLP)
to interact with users and provide information or assistance.
• Data Analysis and Visualization: Tools like Tableau, Power BI, or
Python libraries (e.g., TensorFlow, PyTorch) that analyze and
visualize large datasets.
• Image and Video Processing: Tools like Adobe Photoshop (AI
features), OpenCV, or DALL·E for image recognition, editing, or
generation.
• Predictive Analytics: Tools like IBM Watson or Salesforce Einstein
that predict trends, customer behavior, or outcomes based on
data.
Examples of AI Tools

AI Tools
66
• Automation Tools: Robotic Process Automation (RPA) tools like
UiPath or Automation Anywhere that automate repetitive tasks.
• Healthcare Diagnostics: AI tools like IBM Watson Health or
PathAI that assist in diagnosing diseases or analyzing medical
images.
• Content Creation: Tools like Jasper AI or Grammarly that
generate or refine written content using AI.
• Voice and Speech Recognition: Tools like Google Speech-to-Text
or Amazon Transcribe that convert speech into text or analyze
audio data.
Examples of AI Tools

AI Tools
68
https://www.rapidops.com/blog/generative-ai-tools/

AI Tools
69
• What is an AI tool, and how does it utilize technologies like machine learning and natural
language processing?
• What are some examples of generative AI applications, and how do they create new content?
• How has the journey to generative AI evolved from 2010 to 2022, particularly in natural
language understanding?
• What are some common AI tools used for data analysis, visualization, and predictive analytics?
• How do AI tools like ChatGPT, DeepSeek, and Google Assistant use natural language processing
(NLP) to interact with users?
• What are some AI tools used in healthcare diagnostics, and how do they assist in disease
diagnosis?
• What are the most used AI tools in 2024 across categories like automation, chatbots, and
content creation?
• How do AI tools like Midjourney and Adobe Firefly contribute to design and creative processes?
• What role do AI tools play in productivity and task management, as seen in tools like Notion AI
and Taskade?
• How are AI tools like Runway and Pictory transforming video creation and editing processes?

Concerns of Growth of AI Technology
70
With the rapid global expansion of AI utilization across various
aspects of life, numerous concerns have been raised regarding the
advancement of AI technology. These concerns are outlined below.
Concern of Jobs Displacement
• Fears of losing jobs to AI new technologies.
• The opposite debate is that the artificial intelligence will shift the
demand for jobs to other areas related to use of AI.
• Example: AI as an assistant to doctors, not a replacement. AI can
potentially allow a doctor to make better informed medical
diagnoses much quicker than they could traditionally.
• Though AI may indeed threaten some jobs, it will also help create
new jobs that we perhaps cannot even define today.
https://www.ibm.com/think/topics/machine-learning

71
Ethical Issues
• In 2016, a Rembrandt (Dutch Painter, 1606-1669) painting, “the
Next Rembrandt”, was designed by a computer and created by a
3D printer after 351 years Rembrandt death. The resultant paint
could trick any expert. Who can be the designated author?
• In 2019, the Chinese technology company Huawei announced that
an AI algorithm has been able to complete the last two movements
of Symphony No.8, the unfinished composition that Franz Schubert
started in 1822, about 197 years before.
• What could happen when AI has the capacity to create works of art
itself? Who will claim authorship and copy right, an algorithm or a
fraud?
Social Impact
• Potential for AI to exacerbate societal divides and influence human
behavior negatively.
https://www.unesco.org/en/artificial-intelligence/recommendation-ethics/cases

72
Data Privacy
• There are concerns about the context of data privacy, data
protection, and data security. New legislations need to be issued.
Concern of Bias and Discrimination
• How can we safeguard against bias and discrimination when the
training data itself may be generated by biased human processes?
• AI and learning algorithms extrapolate from the data they are given.
• For example, if you train a human detection algorithm and only
show the algorithm images of people with blonde hair, that system
may fail to recognize a user with brown hair. The result is AI with
racial and gender biases.
https://eng.vt.edu/magazine/stories/fall-
2023/ai.html#:~:text=%E2%80%9CUnfortunately%2C%20AI%20may%20have%20its,of%20water%20for%20cooling%20purposes.

73
Large Carbon Footprint for AI
• AI may have its own carbon footprint and negative environmental
impact because it relies heavily on computing at data centers
which consume a large amount of electricity, and also require a
significant amount of water for cooling purposes.
• Machine learning solutions are based on expert knowledge which
require long time of machine learning training.
• Data centers have a high-and-growing carbon footprint. A large
fraction of the jobs users request are either redundant or
misinformed, with no useful or actionable result.
Lack of Regulation
• Insufficient governance and oversight to ensure safe and fair AI
development.

74
Decision Making Freedom
• Today's AI systems influence human decision making at multiple
levels: from viewing habits to purchasing decisions, from
political opinions to social values. There is the ethical concern of
AI being used to manipulate or deceive, given its ability to
generate convincing narratives.
• Of course companies and institutions are free to develop the
algorithms that maximize their profit, but for sure this influence
our decision making.
• AI as a propaganda engine could be used for polarization and
giving users what are preferred politically, technologically, and in
advertisements.

75
1. What are the primary concerns regarding job displacement due to the growth of AI
technology?
2. How can AI act as an assistant rather than a replacement in professions like healthcare?
3. What ethical issues arise when AI creates works of art, such as paintings or music
compositions?
4. How might AI exacerbate societal divides and influence human behavior negatively?
5. What are the key concerns related to data privacy and security in the context of AI
development?
6. How can bias and discrimination in AI systems be addressed, especially when training data is
biased?
7. What is the environmental impact of AI, particularly in terms of carbon footprint and resource
consumption?
8. Why is there a need for better regulation and governance in the development and
deployment of AI?
9. How does AI influence human decision-making, and what are the ethical concerns associated
with this influence?
10. What are the potential risks of AI being used as a propaganda engine for political or
commercial purposes?

References
References
76
The following references are in addition to others indicated in footnotes of some
slides.
1. https://www.acuitykp.com/blog/staying-ahead-of-the-curve-with-microsoft-
copilot-generative-ai/
2. https://chat.deepseek.com/
3. https://ellow.io/components-of-ai
4. https://en.wikipedia.org/wiki/Sigmoid_function
5. https://gemini.google.com/app
6. https://juliety.com/ai-statistics
7. https://medium.com/towards-data-science/first-neural-network-for-
beginners-explained-with-code-4cfd37e06eaf
8. https://www.geeksforgeeks.org/fuzzy-logic-introduction/
9. https://www.geeksforgeeks.org/neural-networks-a-beginners-guide/

References
References (… continued)
77
10. https://www.geeksforgeeks.org/top-10-branches-of-artificial-intelligence/
11. https://www.gartner.com/en/topics/generative-ai
12. https://www.ibm.com/think/topics/artificial-intelligence
13. https://www.ibm.com/think/topics/machine-learning
14. https://www.rapidops.com/blog/generative-ai-tools/
15. https://www.soci.ai/knowledge-articles/branches-of-artificial-intelligence/
16. https://www.tableau.com/data-insights/ai/history
17. https://www.techtarget.com/searchenterpriseai/definition/natural-language-
understanding-NLU
18. https://www.unesco.org/en/artificial-intelligence/recommendation-
ethics/cases

Discussion
78
http://www.camdenmiracleleague.com/wp-content/uploads/2013/01/thank-you.jpg

T32 AI Trends, Technologies, & Transformations 250323.pdf

More Related Content

Similar to T32 AI Trends, Technologies, & Transformations 250323.pdf (20)

More from Moustafa M Elsayed (20)

Recently uploaded (20)

T32 AI Trends, Technologies, & Transformations 250323.pdf