Emerging
Information Systems
Technologies
Module 2: Section 2
Prof. Jyoti P. Das
Artificial
Intelligence
26-08-2025
Prof. Jyoti P. Das
AI … what is it
In many ways, AI is trying to mimic
Artificial intelligence (AI) is human cognition, such as language
technology that enables computers & speech, reasoning, problem
and machines to simulate human solving, planning etc.
learning, comprehension, problem Now when people talk about AI,
solving, decision making, creativity they are very likely referring to
and autonomy. Machine Learning or Deep Learning
➢Applications and devices equipped (a subset of Machine Learning).
with AI can see and identify objects. Machine learning is the ability for a
➢ They can understand and respond to computer to learn and improve
human language. from experience, like a human,
without being explicitly
➢They can learn from new information programmed for every single task
and experience.
Deep learning is a method that
➢They can make detailed teaches computers to process data
recommendations to users and in a way inspired by the human
experts. Today, most AI researchers, practitioners and brain. Deep learning models can
most AI-related headlines are focused on
➢They can act independently, replacing breakthroughs in generative AI (gen AI), a
recognize complex pictures, text,
the need for human intelligence or technology that can create original text, images, sounds, and other data patterns to
intervention (a classic example being a video and other content. produce accurate insights and
self-driving car). predictions
Prof. Jyoti P. Das
AI … a derivative concept
AI is as a series of nested or derivative concepts that have emerged over more than 70 years
Prof. Jyoti P. Das
AI … history
Prof. Jyoti P. Das
AI … Reasons why industry started
adopting AI
Reasons industries start adopting AI
To curb pain points To provide unique offerings To retain increasing
crowd of customers
Prof. Jyoti P. Das
AI … Reach deep into organizations
to affect end-to-end workflow
Prof. Jyoti P. Das
AI …Types
The idea behind machine learning s Neural networks are computer Natural language processing is often The use of robots dates back
to use sample data to train systems designed to imitate the confused with simple speech hundreds of years, but machines
computer programs to recognize neurons in a brain. recognition. that could assist people without
patterns based on algorithms. The networks are designed to It certainly includes the ability to actual human involvement are
learn via processing examples, understand speech, but it involves considerably newer.
Machine learning programs allow more than simple audio-to-text (and
using a well-understood input and One of the first autonomous robots
computers to do things without vice versa) translations.
a predicted result. was installed in 1961 and used to lift
being programmed. The efficiency of
As the network gains experience, One of the goals of natural language hot pieces of metal from a machine.
machine learning generally depends processing is the design of a computer
its output mirrors the targeted Since then, robots have become far
on the algorithm to fully understand and analyze
result more common. They’re used in
documents
applications ranging from space
exploration to vacuuming your
carpet.
Prof. Jyoti P. Das
AI …Different branches
Evolutionary Computation Expert System
Machine Learning
Is a family of algorithms for global optimization Systems use databases of expert knowledge
Is teaching computers to learn from
inspired by biological evolution/ nature (example: to offer advice or make decisions
data, like humans do, without being
explicitly programmed for every task. genetic algorithm)
Natural Language
Fuzzy Logics Robotics
Processing
Is the field that deals with designing, Is a branch of artificial intelligence that
Is an approach to computing based on
building, and using robots helps computers understand, interpret and
"degrees of truth" rather than the usual
manipulate human language.
"true or false" (1 or 0) Traditional logic
Prof. Jyoti P. Das
AI … next evolution Machine
Learning
Directly underneath AI, we
have machine learning,
Breakdown of Machine Learning
which involves creating
models by training an Subset of AI - Machine learning is a specific
algorithm to make area within the broader field of artificial
predictions or decisions intelligence (AI).
based on data. Learning from Data - Instead of being
programmed with specific instructions, ML
It encompasses a broad systems learn by analyzing data and
range of techniques that identifying patterns.
enable computers to learn
Algorithms and Models - ML uses
from and make inferences
algorithms (sets of rules) that are trained on
based on data without being data to make predictions or decisions. The
explicitly programmed for trained algorithms are called machine
specific tasks. learning models.
Essentially, ML enables Improved Performance - As ML models
machines to learn from are exposed to more data, they learn and
experience, just like humans improve their ability to perform tasks, such as
identifying objects, predicting outcomes, or
do, but through algorithms
making recommendations.
and data analysis.
In essence, machine learning automates the process of learning and improvement, allowing computers to become more effective at tasks
over time without constant human intervention.
Prof. Jyoti P. Das
AI … next evolution Machine
Learning
Neural Network
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.
Supervised Learning
The simplest form of machine learning is called supervised learning ,
which involves the use of labeled data sets to train algorithms to
classify data or predict outcomes accurately.
In supervised learning, humans pair each training example with an
output label. The goal is for the model to learn the mapping between
inputs and outputs in the training data, so it can predict the labels of
new, unseen data.
Prof. Jyoti P. Das
AI vs Machine Learning
Artificial Intelligence
Artificial Intelligence (AI) is Goal -To create machines that can mimic
a broad field focused on human intelligence.
creating machines that can Scope - Encompasses a wide range of
perform tasks typically techniques and approaches to achieve
requiring human intelligent behavior.
intelligence, like reasoning, Examples - Virtual assistants (like Siri or Alexa),
recommendation systems, fraud detection,
problem-solving, and
autonomous vehicles, and more.
learning.
Machine learning (ML), is a Machine Learning (ML)
subset of AI that focuses on Goal - To enable machines to learn from data
and improve their performance on specific
developing algorithms that tasks without being explicitly programmed.
allow machines to learn
Approach - Uses algorithms to analyze data,
from data without explicit identify patterns, and make predictions or
programming. decisions.
ML is a specific approach to Subset of AI - ML is a specific method or
technique for achieving AI.
achieving AI, using data to
train models that can then In essence, machine learning is a powerful tool that Examples: Image recognition, natural
language processing, spam filtering,
perform tasks helps us build AI systems. It allows machines to learn
predictive maintenance.
autonomously. from data and adapt to new situations, making them
more intelligent and capable.
Prof. Jyoti P. Das
AI … Deep Learning
Deep learning is a branch of artificial Key Concepts
intelligence (AI) that uses algorithms Artificial Neural Networks (ANNs) -
inspired by how the human brain Deep learning models are based on ANNs,
works. which are composed of interconnected nodes
(neurons) organized in layers.
It helps computers learn from large
Multiple Layers - Deep learning
amounts of data and make smart
distinguishes itself by having multiple hidden
decisions layers within the neural network, allowing it
Deep learning is a subset of machine to learn increasingly complex representations
learning that uses multilayered neural of data.
networks, called deep neural Automatic Feature Extraction: Unlike
networks, that more closely simulate traditional machine learning, deep learning
the complex decision-making power can automatically extract relevant features
from raw data, reducing the need for manual
of the human brain.
feature engineering.
Deep neural networks include an
Data-Driven Learning - Deep learning
input layer, at least three but usually models learn from data, adjusting their
hundreds of hidden layers, and an internal parameters to optimize performance
output layer, unlike neural networks on a specific task.
used in classic machine learning High-Dimensional Data - Deep learning
models, which usually have only one excels at handling large and complex datasets,
or two hidden layers. including images, text, and audio.
Prof. Jyoti P. Das
AI … Deep Learning
These multiple layers enable unsupervised
learning , they can automate the
extraction of features from large,
unlabeled and unstructured data sets, and
make their own predictions about what
the data represents.
Because deep learning doesn’t require
human intervention, it enables machine
learning at a tremendous scale.
It is well suited to natural language
processing (NPL), computer vision, and
other tasks that involve the fast, accurate
identification complex patterns and
relationships in large amounts of data.
Some form of deep learning powers most
of the artificial intelligence (AI)
applications in our lives today.
In a deep neural network, multiple layers
of nodes can extract meaning and
relationships from large volumes of
unstructured, unlabeled data.
Prof. Jyoti P. Das
AI … Generative AI
Generative AI is a type of Key Characteristics:
artificial intelligence that
Creation of new content - Generative
creates new content, such
AI doesn't just analyze data; it generates
as text, images, music, and entirely new content that didn't exist
videos, based on patterns before.
learned from existing data.
Learning from data - It learns from
Unlike traditional AI that vast amounts of data to understand
analyzes or categorizes patterns and structures, allowing it to
data, Generative AI focuses generate similar but unique content.
on producing original Variety of outputs - Generative AI can
content by mimicking produce various types of content,
human creativity. including text (like articles, stories, code),
images, music, videos, and more.
This is achieved by training
models on large datasets Use of prompts - Users can interact
with generative AI models through
and using algorithms to prompts, which are essentially requests or
generate new material instructions, to guide the AI in creating
based on learned patterns. specific content.
Generative AI has the potential to revolutionize various industries by automating tasks, enhancing creativity, and enabling new forms
of content creation. It can be used in customer service, content creation, software development, and many other fields.
Prof. Jyoti P. Das
AI … how does Gen AI work
Generative AI Generative AI operates in three phases:
works by learning Training, to create a foundation model - Generative AI
patterns from begins with a "foundation model"; a deep learning model that
serves as the basis for multiple different types of generative AI
massive datasets applications.
and then using To create a foundation model, practitioners train a deep learning
those patterns to algorithm on huge volumes of relevant raw, unstructured,
create new, unlabeled data, such as terabytes or petabytes of data text or
original content, images or video from the internet.
The most common foundation models today are Large Language
such as text,
Models (LLMs) created for text generation applications.
images, or code.
Tuning, to adapt the model to a specific application
It leverages - next, the model is tuned to a specific content generation task. This
techniques like can be done in various ways, including:
neural networks Fine-tuning, which involves feeding the model application-specific
and deep labeled data, questions or prompts the application is likely to
receive, and corresponding correct answers in the wanted format.
learning to Reinforcement learning with human feedback (RLHF), in which
analyze the data human users evaluate the accuracy or relevance of model outputs
and generate so that the model can improve itself.
novel outputs that Generation, evaluation and more tuning, to improve
resemble the accuracy - developers and users regularly assess the outputs of
training data. their generative AI apps, and further tune the model even as often
as once a week for greater accuracy or relevance.
Prof. Jyoti P. Das
AI … Benefits
AI offers numerous Automation of repetitive tasks - AI can automate routine,
benefits across various repetitive and often tedious tasks including digital tasks such as
data collection, entering and preprocessing, and physical tasks
sectors by boosting such as warehouse stock-picking and manufacturing processes.
productivity, enhancing This automation frees to work on higher value, more creative
decision-making, and work.
reducing costs through Enhanced decision-making - Whether used for decision
support or for fully automated decision-making, AI enables faster,
automation. more accurate predictions and reliable data-driven decisions.,
It minimizes human Fewer human errors - AI can reduce human errors in various
error, improves ways, from guiding people through the proper steps of a process,
to flagging potential errors before they occur, and fully
accuracy, and facilitates automating processes without human intervention. This is
faster data analysis, especially important in industries such as healthcare where, for
leading to more efficient example, AI-guided surgical robotics enable consistent precision.
operations and Round-the-clock availability and consistency - AI is always on,
innovative solutions. available around the clock, and delivers consistent performance
every time. Tools such as AI chatbots or virtual assistants can
It can also enhance lighten staffing demands for customer service or support.
personalized Reduce physical risk - By automating dangerous work such as
experiences and animal control, handling explosives, performing tasks in deep
ocean water, high altitudes or in outer space, AI can eliminate the
contribute to solving need to put human workers at risk of injury or worse. While they
complex global have yet to be perfected, self-driving cars and other vehicles offer
challenges the potential to reduce the risk of injury to passengers.
Prof. Jyoti P. Das
AI … Use cases
Customer experience, service and support
Companies can implement AI-powered chatbots and virtual assistants to handle customer inquiries, support tickets and more. These
tools use (NLP) and generative AI capabilities to understand and respond to customer questions about order status, product details
and return policies. Chatbots and virtual assistants enable always-on support, provide faster answers to frequently asked questions
(FAQs), free human agents to focus on higher-level tasks, and give customers faster, more consistent service.
Machine learning and deep learning algorithms can analyze transaction patterns and flag anomalies, such as unusual spending or
login locations, that indicate fraudulent transactions. This enables organizations to respond more quickly to potential fraud and limit
its impact, giving themselves and customers greater peace of mind.
Personalized marketing
Retailers, banks and other customer-facing companies can use AI to create personalized customer experiences and marketing
campaigns that delight customers, improve sales and prevent churn. Based on data from customer purchase history and behaviors,
deep learning algorithms can recommend products and services customers are likely to want, and even generate personalized copy
and special offers for individual customers in real time.
Fraud detection
Machine learning and deep learning algorithms can analyze transaction patterns and flag anomalies, such as unusual spending or login
locations, that indicate fraudulent transactions. This enables organizations to respond more quickly to potential fraud and limit its
impact, giving themselves and customers greater peace of mind.
Prof. Jyoti P. Das
AI … Use cases (2)
Human resources and recruitment
AI-driven recruitment platforms can streamline hiring by screening resumes, matching candidates with job descriptions, and even
conducting preliminary interviews using video analysis. These and other tools can dramatically reduce the mountain of administrative
paperwork associated with fielding a large volume of candidates. It can also reduce response times and time-to-hire, improving the
experience for candidates whether they get the job or not.
Application development and modernization
Generative AI code generation tools and automation tools can streamline repetitive coding tasks associated with application
development, and accelerate the migration and modernization (reformatting and re-platforming) of legacy applications at scale.
These tools can speed up tasks, help ensure code consistency and reduce errors.
Personalized marketing
Predictive maintenance
Machine learning models can analyze data from sensors, Internet of Things (IoT) devices and operational technology (OT) to
forecast when maintenance will be required and predict equipment failures before they occur. AI-powered preventive maintenance
helps prevent downtime and enables you to stay ahead of supply chain issues before they affect the bottom line.
Prof. Jyoti P. Das
AI … challenges and risks
AI faces significant
challenges and risks,
including bias, job
displacement, data
privacy concerns, security
vulnerabilities, and the
potential for misuse like
deepfakes and social
manipulation.
Ethical considerations,
such as algorithmic
bias and lack of
transparency, also pose
risks.
Furthermore, the rapid
development of AI
necessitates careful
attention to regulations
and accountability to
mitigate potential harm
Prof. Jyoti P. Das
AI … challenges and risks
Data risks
AI systems rely on data sets that might be vulnerable to data poisoning, data
tampering, data bias or cyber attacks that can lead to data breaches.
Organizations can mitigate these risks by protecting data integrity and
implementing security and availability throughout the entire AI lifecycle, from
development to training and deployment and post deployment.
Model risks
Threat actors can target AI models for theft, reverse engineering or
unauthorized manipulation. Attackers might compromise a model’s integrity
by tampering with its architecture, weights or parameters; the core
components that determine a model’s behavior, accuracy and performance.
Operational risks
Like all technologies, models are susceptible to operational risks such as
model drift, bias and breakdowns in the governance structure. Left
unaddressed, these risks can lead to system failures and cybersecurity
vulnerabilities that threat actors can use
Ethics and legal risks
If organizations don’t prioritize safety and ethics when developing and
deploying AI systems, they risk committing privacy violations and producing
biased outcomes. For example, biased training data , used for hiring
decisions might reinforce gender or racial stereotypes and create AI models
that favor certain demographic groups over others.
Prof. Jyoti P. Das
AI… Framework for ethical AI
governance
Prof. Jyoti P. Das
AI… ethics and governance
AI ethics and
governance refers to the
frameworks, principles,
and practices that guide
the responsible
development and
deployment of artificial
intelligence.
It encompasses
ensuring AI systems are
fair, transparent,
accountable, and
beneficial to society.
This involves addressing
potential risks like bias,
privacy violations, and
job displacement, while
maximizing the positive
impact of AI.
Prof. Jyoti P. Das
AI … what can it do …
AI in farming AI to Generate Human Faces
Smart Paintbrush to draw Photorealistic Sceneries Google DeepMind's Deep Q-learning playing Atari Breakout
Prof. Jyoti P. Das