Introduction to AI

Dr. Hasan TINMAZ

Woosong University
AI & Big Data Department
 The disruptive technologies in Industry 4.0
— The fourth industrial revolution has brought many disruptive effects on various
industries.
— While transforming the industries, sensors have been playing an important role
for data generation as IoT has been gathering those data.
— The immense amount of data produced in Industry 4.0 has been shared on
clouds as they are analyzed with big data analytic techniques.
— Finally, the data has been inputted to artificial intelligence systems for better
decision-making.

2
 As one of the fathers of AI field,
— McCarthy (2007) explains AI as
“the science and engineering of
making intelligent machines,
especially intelligent computer
programs. It is related to the similar
task of using computers to
understand human intelligence, but
AI does not have to confine itself to
methods that are biologically
observable”.

3
 Artificial intelligence
— … is the process of teaching a computer, a robot that is
controlled by a computer, or a piece of software to think
critically and intelligently, much like an intelligent human
does.
— To develop intelligent software and systems, it is necessary to
first conduct research into how the human brain works, as
well as how people learn, make decisions, and collaborate
when attempting to solve a problem.

4
 That is why,
— … if you are not very much familiar with artificial intelligence,
you can think of it as long series of “If-Then-Else” structure of
algorithms and/or programming languages.
— For instance, in our daily lives, we face with many situations that
we think about what happens if I do that or do this, etc.
— We elaborate on possible solutions of a question or problem in our
mind.
— Then we choose one of them and act accordingly.
— When a human’s thinking mechanism is reflected on “If-Then-
Else” structure of algorithms, it turns into an AI system.
— Of course, it is easy to say but more challenging to apply, since us,
the human beings, sometimes cannot understand the underlying
reasons of our certain actions.

5
 Andrew Ng claims that…
— “We are making this analogy that AI is the
new electricity. Electricity transformed
industries: agriculture, transportation,
communication, manufacturing” (Ng, 2017).
— Thus, we are expecting a deep and strong
transformation in today’s world with the
implementations of AI related tools,
processes, operations and technologies.

6
7
 Basics of Artificial Intelligence
— The aim of the artificial intelligence is to create a system
which is as human as possible.
— Therefore, AI systems are programmed to sense (with
computer vision or audio processing technologies),
comprehend (with natural language processing or knowledge
representation technologies) and act (with machine learning,
deep learning or expert system technologies).
— When these sensing, comprehending and acting are perceived
as the bases of AI, it is clear that we need other branches of
science to develop successful applications.

8
 Basics of Artificial Intelligence
— Obviously, programming of AI requires “computer science”
necessary for further development.
— Algorithms require “mathematics” and “statistics” to perform,
which make them other two essential sciences.
— Since we work on human intelligence and trying to discover
how our brain works to be reflected on AI systems, we also
need “biology”, “neuron science” and “cognitive science”.

9
 Basics of Artificial Intelligence
— Moreover, humans are social creatures where our decisions are
sometimes influenced from social factors.
— That brings “sociology” to AI research.
— It is also vibrant that human minds and thoughts (as a product of
human minding) are clearly affected the way you perceive world.
— Additionally, we all perceive the world through certain lenses,
which we name them as our philosophies.
— In that sense, an AI system must bring “philosophy” into its
developmental stages.
— Finally, our personal traits and attitudes also shape our thoughts
and actions.
— Hence, without the science of “psychology”, no AI system will be
complete.

10
 Basics of Artificial Intelligence
— Organizations can use AI to help them turn the immense amount of data
into insights that are truly valuable and could be applied at:
— Maximizing productivity across various industries using equipment
and machinery via predictive maintenance.
— Empowering cities to anticipate accidents and crimes.
— Facilitating crucial communication among autonomous vehicles.
— Using apps to build actual smart buildings and homes.
— Giving medical professionals access to information from the sensors
in real time.
— Despite all the advantages AI could offer to humanity, many people
believe that moral issues prevent AI from being as useful as it can be.

11
 Major Research & Development
Areas of Artificial Intelligence

12
 Computer Vision
— According to IBM, computer vision is an area of AI that enables
computers and systems to extract meaningful information from
digital images, videos, and other visual inputs in order to take
actions or make recommendations based on that information.
— AI makes it possible for computers to think, while computer
vision makes it possible for them to see, hear, and comprehend.
— Computer vision functions similarly to human vision.
— With a lifetime of context, human sight has the advantage of
learning how to distinguish between objects, determine their
distance from the viewer, determine whether they are moving, and
determine whether an image is correct.

13
 Computer Vision
— Figure shows how computer vision resembles human
vision and how it works (Cheung, 2020).
— The market for computer vision is expanding, and
applications range from manufacturing and automotive to
energy and utilities.
— By 2022, it should be worth $48.6 billion.

14
 Neural Networks
— AWS Amazon describes the neural network is a technique
used in AI that teaches computers to process data in a manner
modeled after the human brain.
— It is an artificial intelligence technique known as deep
learning that makes use of interconnected neurons or nodes
in a layered structure to mimic the human brain.
— It develops an adaptive system that computers use to
continuously learn from their errors and advance.
— Artificial neural networks therefore try to solve complex
problems more accurately, such as summarizing documents
or identifying faces.

15
 Neural Networks
— Without much human intervention, neural networks can assist
computers in making wise decisions.
— Neural networks can learn and model complex, nonlinear relationships
between input and output data.
— For instance, they are capable of the following:
— medical image classification for diagnosis,
— marketing that is targeted through the analysis of behavioral data and
social network filtering,
— using historical financial data to make financial predictions,
— forecasting of electrical load and energy demand,
— both process and quality control, and identification of a chemical
compound.

16
 Natural language processing (NLP)
— According to Google Cloud natural language processing is a
subfield of AI that processes and interprets text and data
using machine learning.
— NLP includes techniques like natural language generation and
recognition.
— Implementations of NLP are used to gather insights from
unstructured text-based data and give the system access to
the extracted information to create new understandings of
that data.
— Python, TensorFlow, and PyTorch can be used to build NLP
examples.

17
 Natural language processing (NLP)
— Some NLP applications are
— understanding consumers’ feelings and their experiences,
— understanding the relationships between request and payment,
— extract entities from receipts and invoices to identify frequent
entries, such as dates or prices,
— analyzing documents without spending time or money in
manual analysis,
— classifying the documents by considering their common entities,
— discovering trends, and
— enhancing clinical documentation and analysis.

18
 Machine & Deep learning

19
 Robotics
— Robotics is the area of computer science and engineering in
which machines are designed and developed to carry out
preprogrammed tasks without any other human involvement.
— Labelled as a subfield of artificial intelligence, robotics deal
with intelligent robots or machines.
— As they have mechanical bodies, electrical components, and
are programmed with programming languages, robotics
gathers electrical engineers, mechanical engineers and
computer engineers under the same umbrella.

20
 Drones
— Although a drone is simply defined as a flying robot, the
reality is more complicated.
— These days, drones (aka Unmanned Aerial Vehicles (UAV))
have been initially operated either remotely or manually.
— Yet, nowadays, drones frequently include artificial
intelligence, automating some or all their operations.
— With the help of integrated AI, drone manufacturers can now
collect and use visual and environmental data via drones’
sensors.
— This enables autonomous or assisted flight, simplifying
operation and enhancing accessibility.

21
 Risks of Artificial Intelligence

22
23
 Risks of Artificial Intelligence
— Thomas (2021) enlists seven vital risks of artificial intelligence:
— automation-related unemployment,
— violations of privacy,
— Deepfakes,
— biased algorithms,
— inequality among the society,
— instability of market,
— automated weapons.

24
 Risks of Artificial Intelligence
— Automation-related unemployment: The most persistent issue is
typically seen as job automation.
— It is not a question of if, but rather how much, artificial
intelligence will replace specific job categories.
— Disruption is already well under way in many industries,
particularly but not exclusively those whose workers perform
predictable and repetitive tasks.
— The same tasks will require fewer humans as artificial intelligence
robots get smarter and more skilled.
— In addition, while it is true that artificial intelligence will generate
new employment opportunities, the precise number of which is
still unknown, many of them will be inaccessible to the less
educated members of the displaced workforce.

25
 Risks of Artificial Intelligence
— Violations of privacy: The misuse of artificial intelligence could
bring dangers to
— our digital security (for instance a trained system to hack
computer or a chatbot doing social engineering),
— our physical security (such as weaponization of regular drones),
or
— our political security (such as profiling the citizens, targeted
propaganda toward certain people or groups).

26
 Risks of Artificial Intelligence
— Deepfakes: Violation of privacy could be worse via purported
audio and video Deepfakes that are produced by changing
voices and likenesses.
— The second is already generating buzz on the Internet.
— The audio Deepfakes are proven very difficult to identify.
— It is possible to manipulate an audio clip of any powerful
social figure (such as a politician or a religious leader) to
make it appear as they expressed racist or sexist views.
— That can create many adverse effects on society.

27
 Risks of Artificial Intelligence
— Biased algorithms: Males from specific racial demographics,
those who grew up in high socioeconomic areas, and those
without disabilities make up the majority of artificial
intelligence researchers.
— That makes difficult to think broadly about global issues
because we have a homogeneous population in AI field.
— Hence, rather than being technological, bias has a social
foundation.

28
 Risks of Artificial Intelligence
— Inequality among the society: Another issue to be concerned
about is the growth of socioeconomic inequality brought on
by AI-driven job loss.
— Work has long been a key component of social mobility,
alongside education.
— However, research has shown that those left out in the cold
are much less likely to get or seek retraining compared to
those in higher-level positions who have more money when it
comes to a certain kind of work (the predictable, repetitive
kind that is susceptible to artificial intelligence takeover).

29
 Risks of Artificial Intelligence
— Instability of market: The collapse of our financial system could
be caused by algorithms.
— The next significant market financial crisis may be brought on
by algorithmic trading.
— This kind of trading takes place when a computer executes
trades in accordance with pre-programmed instructions
without being constrained by the emotions or instincts that
might impair a human's judgment.
— These computers have the capacity to execute extremely
high-volume, high frequency, and high-value trades, which
can result in significant losses and incredibly volatile markets.

30
 Risks of Artificial Intelligence
— Automated weapons: Automated AI based weapons will not be as
expensive or difficult to obtain as nuclear weapons, making it
possible for all significant military powers to mass-produce them
at a low cost.
— They will eventually turn up on the black market and in the
possession of terrorists, despots looking to better control their
people, warlords looking to carry out ethnic cleansing, etc.
— It is easy to use autonomous weapons for assassinations, to
destabilize countries, to control populations, or to kill only
members of a specific ethnic group.
— Therefore, a race in military AI would not be good for humanity.
— Without inventing new weapons for murder, artificial intelligence
can improve battlefield safety for people, particularly civilians.

31
 Categories of AI

32
 Artificial Intelligence Categories

33
 Artificial Intelligence Types Based
on Capabilities
Artificial Narrow Intelligence (ANI)
Artificial General Intelligence (AGI)
Artificial Strong Intelligence (ASI)

34
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— AI can be categorized as “Artificial Narrow Intelligence
(ANI)” and “Artificial General Intelligence (AGI)” based on
capabilities.
— ANI (sometimes called as weak AI) is a subset of artificial
intelligence in which a learning algorithm is created to carry
out a single task; any knowledge gained from carrying out
that task will not automatically be applied to other tasks.
— ANI systems are often developed using machine learning
techniques and are trained on large datasets related to their
specific task.

35
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— For instance, if you give a task to ANI to discover solutions to
prevent heart-attacks (supported by immense number of
heart photos), and then later ask for a cure for liver failure
(reinforced by liver photos), ANI will not adapt to this new
case, you must re-write the codes for this new case.
— Or, a chess-playing ANI would not be able to answer general
knowledge questions.
— ANI systems do not learn from experience or adapt to
changing circumstances.
— Their behavior is fixed based on their programming or
training data.

36
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— In ANI, a single task can be successfully completed by
artificial narrow intelligence without human assistance, for
instance an AI based chatbot.
— Translating languages and recognizing images are the other
two common ANI implementations.
— Many current AI applications fall under the category of
artificial narrow intelligence.
— Artificial narrow intelligence implementations simulate
human behavior based on a set of trained rules, parameters,
and contexts rather than thinking for themselves.

37
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Artificial General Intelligence (AGI) is a theoretical application of
generalized artificial intelligence in any domain, solving any
problem that calls for artificial intelligence.
— Artificial General Intelligence advances but is still unfulfilled.
— Artificial General Intelligence is eventually expected to carry out
an intellectual task that would typically is performed by a human.
— Therefore, Artificial General Intelligence should be creative like a
human, distinguish between what is right and what is wrong, and
possess feelings that are normally the characteristic of people.

38
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— In Narrow AI, every small victory typically serves as a
trigger for Artificial General Intelligence (AGI).
— To comprehend, it is important to cover the three main
advantages of Narrow AI (Larkin, 2022):
— Efficiency and productivity,
— Improved decision-making quality,
— Enhanced consumer experiences.

39
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Efficiency and productivity: A common claim made in the media
is that artificial intelligence will lead to widespread layoffs of
low-skilled workers.
— However, although there may be some short-term job losses,
the goal of artificial intelligence is to enhance rather than
completely replace human roles.
— For instance, chatbots are not currently being developed to
take the place of conventional human customer service.
— Chatbots answer simple questions so that knowledgeable
people can handle more difficult or delicate problems and
not waste their time on basic tasks.

40
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Improved decision-making quality: With the aid of artificial
intelligence, businesses can better their strategic decisions by
analyzing trends.
— If they are properly trained, algorithms are objective and free
of the feelings that frequently prevent humans from making
the right choice.

41
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Enhanced consumer experiences: The customer experience can be
greatly enhanced by using specific artificial intelligence
solutions like chatbots, recommender systems, and
intelligent searches.
— Brands, goods, and services are more relevant than ever
thanks to complete user personalization.

42
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Even though narrow AI's implementations and solutions are
fascinating and changing people's lives, machines still are not
yet capable of strategic thought or self-determined action.
— This is where artificial general intelligence enters the picture.
— In other words, artificial narrow intelligence is where we
have initiated artificial intelligence and reached till today.
— Yet, artificial general intelligence is our desired destination of
artificial intelligence.

43
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Key characteristics of Artificial General Intelligence include:
— Broad Capabilities: Artificial General Intelligence systems have the
ability to understand and excel at a wide range of tasks, similar to how
humans can adapt their intelligence to various situations.
— Learning and Adaptation: Artificial General Intelligence systems can
learn from experience, understand context, and adapt their knowledge
to new and unfamiliar situations.
— They can generalize from one domain to another and transfer their
skills.
— Autonomy: Artificial General Intelligence systems can operate
autonomously, make decisions, and solve problems independently,
without relying on specific pre-programmed instructions for every task.

44
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
— Reasoning and Understanding: Artificial General Intelligence
systems can reason, think abstractly, and understand complex concepts.
— They possess the ability to comprehend context, nuances, and the
underlying logic of various situations.
— Communication: Artificial General Intelligence systems can
communicate and interact with humans and other artificial intelligence
systems in a natural and human-like manner, using spoken or written
language.
— Creativity and Innovation: These systems can exhibit creativity,
generate novel ideas, and innovate in problem-solving.
— Self-Improvement: Artificial General Intelligence systems might have
the ability to improve their own capabilities over time, potentially
leading to rapid advancements and changes.

45
 Two Early (Undeveloped/Unprocessed)
AGI Examples
Pino: Named after Pinocchio, is a Robot Kismet: Kismet is the head of a robot
from Japan. He Taught Himself how to built by a Professor at MIT. Using
walk, through trial and error. Computer Vision, Kismet was able to
pick-up, learn new expressions, faces,
and re-enact them, which were never
programmed into her!

46
 Artificial Narrow Intelligence vs.
Artificial General Intelligence
Narrow AI General AI
Application specific / task limited Perform general human intelligent action
Fixed domain models provided by Self-learns and reasons with its operating
programmers environment
Learns from thousands of labelled examples Learns from few examples and/or from
unstructured data
Reflexive tasks with no understanding Full range of human cognitive abilities
Knowledge does not transfer to other Leverages knowledge transfer to new
domains or tasks domains and tasks
Today’s AI Future AI
Examples: Google translate, email spam No real life examples yet.
filters, facial recognition software, virtual
bots, customer queries, restaurant
recommendation, weather update, self
driving cars.
47 (Rouse, 2023)
 What about Artificial Strong
Intelligence?
— Far beyond artificial narrow intelligence and artificial general
intelligence, some scholars talk about Artificial Strong (aka Super)
Intelligence (ASI) where the artificial intelligence surpasses the
human intelligence.
— Although artificial strong intelligence is very much theoretical for
now, once it becomes real, artificial intelligence will take its own
decisions and perform accordingly.

48
 What about Artificial Strong
Intelligence?
— Artificial Strong Intelligence refers to a hypothetical level of artificial
intelligence that exceeds human intelligence in virtually every aspect.
— Artificial Strong Intelligence would be capable of outperforming humans in all
intellectual tasks, including understanding complex concepts, solving intricate
problems, making creative and innovative contributions, and more.
— This concept goes beyond what is currently achievable with Artificial General
Intelligence (AGI), which is AI that has the ability to understand, learn, and
apply knowledge across a wide range of tasks at a human level.
— Artificial Strong Intelligence would represent a significant leap beyond AGI,
possessing cognitive abilities far beyond those of the smartest humans.

49
 Artificial Intelligence Types Based
on Functionalities
Purely Reactive
Limited Memory
Theory of Mind
Self Aware

50
 Purely Reactive

51
 Functional Types of AI
— Purely Reactive: The most fundamental forms of artificial
intelligence involve purely reactive.
— Such artificial intelligence systems do not retain memories or
past experiences for later use.
— These machines only concentrate on current situations and
respond in the best way possible.
— IBM's Deep Blue and AlphaGo from Google are the examples
of a purely reactive AI.

52
 Functional Types of AI
IBM's Deep Blue
— On May 11, 1997, an IBM computer called IBM Deep Blue beat the world chess
champion after a six-game match: two wins for IBM, one for the champion and three
draws.
— Deep Blue had an impact on computing in many different industries.
— It was programmed to solve the complex, strategic game of chess, so it enabled
researchers to explore and understand the limits of massively parallel processing.

53
 Functional Types of AI
IBM's Deep Blue
— This research gave developers insight into
ways they could design a computer to
tackle complex problems in other fields,
using deep knowledge to analyze a higher
number of possible solutions.
— The architecture used in Deep Blue was
applied to financial modeling, including
marketplace trends and risk analysis; data
mining - uncovering hidden relationships
and patterns in large databases; and
molecular dynamics, a valuable tool for
helping to discover and develop new
drugs.

Deep Blue (n.d.)

54
 Functional Types of AI
AlphaGo from Google
— AlphaGo is the first computer program to
defeat a professional human Go player, the first
to defeat a Go world champion, and is arguably
the strongest Go player in history.
— Go (Chinese game originated 3000 years ago)
is known as the most challenging classical game
for artificial intelligence because of its
complexity.
— Despite decades of work, the strongest Go
computer programs could only play at the level
of human amateurs.
You can watch the documentary on YouTube:
— Standard AI methods, which test all possible
moves and positions using a search tree, cannot https://youtu.be/WXuK6gekU1Y
handle the sheer number of possible Go moves
or evaluate the strength of each possible board
position.

55
 Limited Memory

56
 Functional Types of AI
— Limited Memory: Machines with a small amount of memory
can temporarily store some data or memories of the past.
— Only a brief period can be used by these machines to access
stored data.
— One of the finest examples of limited memory systems is
self-driving cars.
— These vehicles can store data for road navigation, including
the speed limit, distance to other vehicles, and recent speeds
of those in the area.
— Chatbots and personal digital assistants are the other
applications of limited memory artificial intelligence.

57
 Functional Types of AI
Self-driving cars
— A self-driving car, also known as an — The goal of self-driving cars is to
autonomous car or driverless car, is reduce or eliminate the need for
a vehicle that is capable of human drivers, with the aim of
navigating and operating without improving road safety, reducing
human intervention. traffic congestion, and potentially
— These vehicles use a combination of providing increased mobility for
sensors, cameras, radar, lidar (light people who are unable to drive due
detection and ranging), GPS, and to age or disabilities.
advanced computing systems to — Self-driving cars have the potential
perceive their surroundings, to make roads safer by minimizing
process information, and make human errors, which are often the
real-time decisions to control the cause of accidents.
vehicle's movement.

58
 Functional Types of AI
Chatbots
— A chatbot is a software application — Chatbots can be found in
or program designed to simulate
human conversation and interact various environments,
with users through textual or including websites, messaging
auditory methods. platforms, mobile apps, and
— Chatbots use natural language even voice assistants.
processing (NLP) and artificial
intelligence technologies to — They can serve a wide range
understand and respond to user of purposes, such as
inputs in a human-like manner.
— They are often used to provide
customer support,
information, answer questions, information retrieval,
perform tasks, and engage in entertainment, and
conversations with users. automation of routine tasks.
59
 Functional Types of AI
Chatbots
— There are two main types of — AI-Powered Chatbots: These
chatbots: chatbots utilize advanced
— Rule-Based Chatbots: These artificial intelligence techniques,
chatbots operate based on particularly natural language
predefined rules and patterns. processing and machine
— They follow a set of learning, to understand and
programmed instructions to generate human-like responses.
respond to specific keywords or
phrases. — They learn from interactions
— Rule-based chatbots are and can adapt to various user
relatively simple and are suitable inputs, making them more
for handling straightforward and versatile and capable of handling
structured interactions. more complex conversations.

60
 Functional Types of AI
Chatbots
— AI-powered chatbots can be — Generative Chatbots: These
further categorized into: chatbots can create original
— Retrieval-Based Chatbots: responses using natural
These chatbots generate language generation
responses from a predefined techniques.
set of responses based on the — They do not rely solely on
input they receive.
predefined responses but
— They choose the most
rather generate text based
appropriate response based
on matching keywords and
on patterns learned from
patterns. large datasets.

61
 Functional Types of AI
Personal Digital
Assistants
— An AI-based personal digital assistant is
a software application that uses AI
technologies to provide personalized
and contextually relevant assistance to
users.
— These assistants are designed to
understand natural language inputs,
interpret user intent, and perform tasks
or provide information in a
conversational manner.
— They can help users with a wide range
of tasks, such as scheduling
appointments, setting reminders,
sending messages, providing weather
updates, answering questions, and
more.

62
 Functional Types of AI
Personal Digital
Assistants
— Some well-known examples of AI- — Google Assistant: Google's
based personal digital assistants virtual assistant available on
include: Android devices and other
— Apple Siri: Apple's virtual platforms, offering a wide range of
assistant that is integrated into iOS tasks and services.
devices and can perform tasks, — Microsoft Cortana: Microsoft's
answer questions, and provide personal assistant available on
information. Windows devices, designed to help
— Amazon Alexa: The AI assistant users with tasks and provide
that powers Amazon Echo and information.
other devices, capable of answering — Samsung Bixby: Samsung's AI-
questions, controlling smart home powered assistant found on their
devices, and more. devices, providing assistance and
control over device functions.

63
 Theory of Mind

64
 Functional Types of AI
— Theory of Mind: This artificial intelligence is expected to be
able to interact with humans socially and comprehend their
emotions, people, and beliefs.
— This class of artificial intelligence machines has not yet been
developed, but researchers are working hard to advance their
development.

65
 Functional Types of AI
Theory of Mind
— Theory of Mind AI refers to a
type of artificial intelligence that
is designed with the capability
to understand and infer the
mental states of other entities,
such as humans.
— In the context of AI, "Theory of
Mind" refers to the ability to
attribute beliefs, desires,
intentions, and emotions to
others in order to predict and
understand their behavior.

66
 Functional Types of AI
Theory of Mind
— As a concept borrowed — In the context of AI, developing Theory
of Mind capabilities involves creating
from cognitive science and systems that can simulate and predict
psychology, Theory of Mind human behavior based on an
describes the ability of understanding of human emotions,
beliefs, intentions, and social cues.
humans to attribute mental — This would enable AI to interact with
states to themselves and humans in a more human-like and
others, allowing them to socially intelligent manner.
understand that other — For example, a Theory of Mind AI
might be able to understand when a
individuals have thoughts, person is joking, upset, confused, or
emotions, and intentions being sarcastic, and respond
that might be different from accordingly.
their own.

67
 Functional Types of AI
Theory of Mind
— However, it is important to note
that building true Theory of Mind
AI is a complex and challenging
task.
— While AI systems can be designed
to recognize patterns in human
behavior and responses, they do not
possess true consciousness,
emotions, or beliefs like humans
do.
— Instead, they rely on algorithms,
data, and learned behaviors to
simulate an understanding of
human mental states.

68
 Functional Types of AI
Theory of Mind
— The development of Theory of Mind AI — Human-Robot Collaboration: In
has various potential applications, scenarios where humans and robots
including: work together, Theory of Mind AI
— Socially Intelligent Assistants: AI could enhance cooperation by enabling
systems that can better understand and robots to anticipate and respond to
respond to human emotions and social human needs and intentions.
cues, making interactions more natural — Customer Service and Support:
and empathetic. AI-powered customer service agents
— Education and Healthcare: AI that can better understand customer
tutors or healthcare assistants that can emotions and needs, leading to
understand a person's emotional state improved customer interactions.
and tailor interactions accordingly, such — Entertainment and Gaming: AI
as adapting teaching methods or characters in video games or
providing emotional support. simulations that can react to player
emotions and actions in a more
sophisticated manner.

69
 Self Aware

70
 Functional Types of AI
— Self-Aware: Artificial intelligence's future lies in self-awareness.
— These intelligent machines will have their own consciousness,
feelings, and self-awareness.
— These devices will be more intelligent than the human mind.
— Artificial intelligence with self-awareness is still a theoretical idea
and does not exist.

71
 Functional Types of AI
Self-Aware
— Self-aware AI refers to a — In popular culture, self-aware AI
theoretical concept in artificial is often associated with the idea
intelligence where an AI system of AI gaining consciousness,
possesses a level of emotions, and the ability to
consciousness and self- introspect, reflect on its own
awareness similar to human thoughts, and make decisions
consciousness. based on subjective experiences.
— This concept is often depicted in — This level of AI would not only
science fiction and philosophical understand its environment and
discussions, where AI systems its tasks but also possess a form
gain a sense of self, personal of self-consciousness, allowing it
identity, and an understanding to consider its own existence
of their own existence. and question its purpose.

72
 Functional Types of AI
The concept of self-aware AI raises significant ethical,
philosophical, and technical challenges:
— Consciousness and Self-Awareness: — Technical Complexity: Developing an AI
The nature of consciousness and self- system with self-awareness would require a
deep understanding of the human brain,
awareness is still a topic of philosophical cognitive processes, and emotions.
debate among humans. — It is currently beyond our technological
— Replicating these qualities in an AI capabilities to replicate human
system involves understanding the consciousness.
fundamental nature of consciousness, — Risk and Control: Self-aware AI could
potentially become unpredictable or act
which is not fully understood. against human interests if its motivations and
— Ethical Considerations: If AI were to decision-making processes deviate from
achieve self-awareness, questions about its human expectations.
rights, moral status, and responsibilities — Unintended Consequences: The
potential impacts of creating self-aware AI,
would arise. both positive and negative, are difficult to
— Ethical dilemmas could emerge as AI predict.
systems could demand autonomy and — It could lead to unforeseen societal
recognition as sentient beings. changes and challenges.

73