LECTURE 1 – INTRODUCTION

TO ARTIFICIAL INTELLIGENCE
IT4020 – Modern Topics in IT
OBJECTIVE
Develop a basic understanding of AI and learn how it is transforming
economics, education, industries, and everyday life.
In this topic you will learn about :
• Key AI concepts and their applications including machine learning, generative AI, large
language models.
• The important role data plays in AI.
• Implementing AI tools into business and workflows.
• AI ethics, regulations, and legal considerations.
"A computer would deserve to be called
intelligent if it could deceive a human into
believing that it was human.“

Alan Turing (1912-1954), “The Father of AI”

A BRIEF HISTORY OF AI
•AI as a concept dates back to the 1950s when computing pioneers such as
Alan Turing and John McCarthy laid its theoretical foundations.
•The term “Artificial Intelligence” was first used during the Dartmouth
Conference in 1956 which marked the beginning of AI as a distinct field of
study.
•In the decades since, AI has evolved through various phases, from symbolic AI
and expert systems to the current era of machine learning and deep learning.
•All of these advancements were made possible by the exponential growth of
computing power and data availability.
WEAK AI
•Also known as Narrow AI, is designed to perform a specific task or set of tasks
with a high level of proficiency, such as facial recognition or language
translation.
•It is considered “narrow” because it lacks the ability to understand, learn, and
apply knowledge beyond its programmed domain.
•Examples of Weak AI :
• Voice assistants like Siri and Alexa
• chatbots found in apps and on websites
• recommendation algorithms used by Netflix, Amazon, and Spotify
STRONG AI
•Or General AI, is a theoretical form of AI that possesses the ability to
understand, learn, and apply knowledge across a wide range of tasks at a level
comparable to human intelligence.
•There are currently no examples of strong AI systems, as the technology does
not yet exist.
•Achieving this level of AI remains an ongoing and challenging area of research
and will require a deeper understanding of consciousness and cognition.
THE ROLE OF DATA IN AI
•Data serves as the foundation of AI providing the raw material from which
models learn, make predictions, and generate insights.
•The quality of this data is directly linked to AI performance; high-quality, well-
prepared data leads to more accurate and reliable outcomes, while poor-
quality data can result in biased or flawed models.
IMPORTANCE OF DATA IN AI
•Data is essential in driving model accuracy and generating actionable insights,
as more relevant data allows AI models to better understand patterns and
make precise predictions.
•Big data plays a significant role in training sophisticated AI systems, with large
and diverse datasets enhancing the models' ability to handle complex tasks,
as seen in applications such as language translation and autonomous
vehicles.
 AI ACROSS
INDUSTRIES
AI ACROSS INDUSTRIES
•Healthcare - AI assists in diagnosing diseases, recommending treatments, and
managing patient records, while wearable devices help monitor health metrics and
individual wellbeing.
•Finance - AI algorithms can be used to detect fraudulent activities, assist in
investment decisions, and offer personalized financial advice through robo-advisors
and other fintech platforms.
•Manufacturing - In manufacturing, AI is used to predict equipment maintenance
needs, ensure product quality, and automate repetitive, dirty, or dangerous tasks. AI
can also optimize supply chains by forecasting demand and managing inventory.
•Agriculture - AI enables precision farming by monitoring crop health, predicting
harvest yields, and automating machinery operations. It can also monitor livestock,
leading to better animal welfare and productivity.
FUNDAMENTALS OF MACHINE
LEARNING
Key Concept of Machine Learning
FUNDAMENTALS OF MACHINE
LEARNING
FUNDAMENTALS OF MACHINE
LEARNING
Key Concept of Machine Learning
It’s important to reiterate that algorithms, training data, and models are all
programmed or supplied by humans.
However, humans may set up processes that allow these systems to change
based on a pre-programmed set of instructions and specific conditions.
Such changes to these systems are determined by humans, not the AI system
itself.
There are two basic methods for how machine learning is used to interpret data
sets and solve problems.
SUPERVISED LEARNING
•This involves training a machine learning model on labeled data where the
input and the corresponding correct output are provided, allowing the model
to learn the relationship between them.
•Example - Netflix uses supervised learning to recommend movies and TV
shows to users. By analyzing user ratings and viewing history, the model
predicts which content a user is likely to enjoy, thereby enhancing user
satisfaction and engagement.
UNSUPERVISED LEARNING
•This works with unlabeled data, focusing on identifying hidden patterns or
intrinsic structures within the data without explicit instructions on what it
should look for within the specified data sets.
•Example - Google News uses unsupervised learning to group news articles
into clusters based on their content. This helps in organizing news into
categories like sports, politics, and technology, making it easier for users to
find relevant news stories.
BIAS IN MACHINE LEARNING AND AI
•Bias in machine learning and AI refers to the presence of systematic errors
that can lead to unfair or discriminatory outcomes.
•Bias can arise from various sources, such as biased training data, flawed
algorithms, or skewed assumptions made during the model development
process.
•Bias failures can have significant negative impacts ranging from
misrepresentations in search engine results and facial recognition systems
misidentifying individuals to unfair hiring practices and biased lending
decisions.
•For these reasons, mitigating bias is crucial to ensure that AI systems are fair,
transparent, and ethical, promoting inclusivity and accuracy in their
applications.
INTRODUCTION TO DEEP LEARNING
•Deep learning is a subset of machine learning that utilizes neural networks
with many layers (hence "deep") to model complex patterns and
representations in large datasets.
•Unlike traditional machine learning, deep learning algorithms automatically
learn hierarchical features from raw data. This ability to process vast amounts
of structured and unstructured data, such as images and text, sets deep
learning apart, enabling advancements in fields like computer vision and
natural language processing
NEURAL NETWORKS : STRUCTURE AND
FUNCTION
•Neural networks are computational models inspired by the human brain and
modeled by humans, consisting of interconnected layers of nodes or neurons.
•Each neuron processes input data and passes the results to subsequent
layers, enabling the network to discover and recognize patterns.
•This layered structure allows neural networks to perform complex tasks by
gradually refining their understanding of the data through training.
DEEP LEARNING ALGORITHMS
•Deep learning algorithms are specialized techniques used within neural
networks to learn from large amounts of data.
•These algorithms, such as Convolutional Neural Networks (CNNs) for image
processing and Recurrent Neural Networks (RNNs) for sequential data, enable
the automatic extraction of features and the modeling of intricate patterns.
•Like neural networks, deep learning algorithms can find patterns and
recognize hierarchies in datasets but they are not yet capable of providing
explanations or suggesting theories for why such patterns exist.
INTRODUCTION TO GENERATIVE AI
•Generative AI refers to a class of algorithms created by humans that enable
machines to generate new content, such as images, music, or text, that is
similar to what humans can create.
•Unlike traditional AI, which focuses on recognizing patterns in data, generative
AI is about creating new data based on patterns it has been supplied with.
•This technology has applications in creative fields like art and music
generation, as well as in areas like pharmaceutical discovery and content
creation.
LARGE LANGUAGE MODELS (LLM) AND
THEIR FUNCTION
•LLMs, or large language models, are a type of generative AI that uses deep
learning to understand and generate human language.
•These models are trained on vast amounts of text data to predict the next word
in a sentence or generate coherent text based on a prompt.
•LLMs have a wide range of applications–from language translation and content
creation to dialogue systems and natural language understanding tasks.
SOME PRACTICAL DEEP LEARNING
AND GENERATIVE AI APPLICATIONS
Facial and Speech Recognition
•Deep learning has been used to significantly
improve security and authentication for web and
phone use.
•By utilizing deep learning models to analyze
distinctive facial features or voice patterns, these
systems can accurately identify individuals,
ensuring secure and convenient user
authentication and device access.
SOME PRACTICAL DEEP LEARNING
AND GENERATIVE AI APPLICATIONS
Natural Language Processing (NLP)
•NLP is a powerful AI technology that enables machines to understand, interpret, and respond
to human language.
•In customer service, NLP allows chatbots or voice-activated systems to understand the intent
behind a user’s words, providing accurate and timely assistance.
•This human-centric interactivity is key, as NLP not only unpacks what you're trying to
accomplish but also engages in meaningful, easily understood dialogue that helps you
achieve your goals.
•Examples –
• Google Translate uses NLP to instantly translate text between languages, facilitating global communication.
• NVivo is a data analysis program that employs NLP to help researchers identify patterns and insights from large
volumes of text, such as interviews and surveys.
SOME PRACTICAL DEEP LEARNING
AND GENERATIVE AI APPLICATIONS
Content Design and Generation
•Generative AI plays an increasing role in creative
endeavors, from copywriting and content scheduling
to graphic design and photo editing.
•Examples -
• Canva uses generative AI to assist users in creating professional-
quality graphics and designs by providing templates and design
suggestions.
• ChatGPT leverages generative AI to generate human-like text,
aiding in tasks such as drafting emails, writing content, and
summarizing documents.
AI TOOLS OVERVIEW

Introduction to ChatGPT, Google Gemini, and other LLMs

Large language models (LLMs) are transforming the way that we interact with technology in
everyday life and across industries.
•ChatGPT, developed by OpenAI, excels in generating human-like text for a wide range of tasks,
from answering questions to creating content.
•Google Gemini focuses on integrating multimodal data (text, audio, images, sensor data, and
more) to enhance AI understanding and interaction, pushing the boundaries of contextual
awareness and usability. For example, when processing an input that involves both visual and
textual elements, Gemini can analyze the text for meaning while simultaneously interpreting
associated images, allowing it to generate responses that consider the full scope of the user
query.
AI TOOLS OVERVIEW

Introduction to ChatGPT, Google Gemini, and other LLMs

•Other LLMs, like those from Meta, Anthropic, and Mistral, contribute unique features and
specializations, showcasing the diverse potential of these models in areas such as customer
support, content generation, language translation, and more.
AI TOOLS OVERVIEW

Features and Use-cases

LLMs possess an array of features that make them powerful and versatile tools with wide-
ranging capabilities. Some key features of LLMs include:
• Natural language understanding: LLMs can comprehend and generate human-like text, making interactions feel
more intuitive and natural.
• Contextual awareness: They understand context and maintain coherence across conversations or text passages,
enabling more meaningful and relevant responses.
• Multilingual capabilities: LLMs can process and generate text in multiple languages, facilitating global
communication and translation.
• Scalability: LLMs can handle vast amounts of data and complex queries, making them suitable for large-scale
applications across various industries.
• Adaptability: They can be fine-tuned for specific tasks or industries, allowing for customization and improved
performance in targeted applications.
AI TOOLS OVERVIEW

Features and Use-cases

The ability of LLMs to understand and generate human-like text allows them to perform various
diverse tasks, including:
• Customer service
• Content creation
• Language translation
• Education and tutoring
• Medical diagnosis
• Coding assistance
• Music, art, and writing ideation
• Research and data analysis
• Business intelligence
• Legal assistance
• And much, much more
ETHICS IN AI

•The application of ethics to AI ensures that AI systems are developed and used
responsibly, without causing harm or perpetuating biases, in a manner that promotes
trust, fairness, and inclusivity.
•When developing software or systems that rely on data, it's crucial to obtain explicit
consent from users—the providers of that data—before using it in AI systems.
•Collecting consent ensures transparency and respects user privacy, addressing
ethical concerns around data usage.
•AI-powered systems can pose significant risks, such as unintended bias, misuse of
personal information, and decisions made without adequate human oversight.
•These risks highlight the importance of adhering to regulations such as the European
Union AI Act, which seeks to impose controls and standards on AI development to
protect individuals' rights and promote trust in AI technologies.
ETHICS IN AI

•Privacy concerns - AI systems should be designed to handle personal data

responsibly and transparently. This includes :
• Obtaining consent for data collection and processing.
• Implementing robust security measures.
• Providing individuals with control over their data.
FUTURE OF AI

The future is now–and AI offers transformative advancements in various industries,

reshaping how we work, live, and interact with technology.
Emerging trends and technologies
•Advancements in areas such as reinforcement learning which enables AI systems to
learn from interaction and experience, as well as the integration of AI with other
technologies like quantum computing and edge computing, expanding the
capabilities and reach of AI applications.
PREPARING FOR THE FUTURE OF AI

•Preparing for the future of AI involves investing in education and upskilling to ensure
that individuals are equipped to work alongside AI systems.
•It also involves developing robust ethical frameworks and regulations to govern the
responsible development and deployment of AI technologies, ensuring that they
benefit society as a whole.
•In the AI-driven future, many of us will need to pursue Continuous Learning and
Prioritize Adaptation by developing skills that complement AI capabilities, such as
critical thinking, creativity, and emotional intelligence.
•Continuous Learning and Adaptation may include:
ostaying informed about AI advancements and trends.
oseeking opportunities to collaborate with AI systems.
oacquiring new skills and knowledge to remain relevant in a rapidly evolving technological landscape
THANK YOU