If you’re reading this book, then you already know that artificial intelligence (AI) is changing the world. To get us started, I decided to sit down for an interview with an AI, ChatGPT (https://chatgpt.com/). ChatGPT can be an AI voice-to-voice chatbot that attempts to recognize and respond to human emotion. Here’s the conversation we had (both the recording and transcript). So, what would you like to chat about next? I am always eager to explore new topics.

	Hey, chat. GPT, how are you doing today?
	Hey, I’m doing great. Thanks for asking. How about you?
	I’m doing well as myself. So, what do you like to be called? Do you have a preferred name?
	I’m glad to hear you’re doing well! You can just call me ChatGPT. Is there anything else you’d like to chat about?
	Of course, as an AI, what do you think about the future of artificial intelligence?
	The future of AI is pretty exciting! We’re likely to see AI becoming even more integrated into daily life, helping with everything from personalized education to advanced healthcare diagnostics. There are also important conversations happening around ethics, safety, and ensuring that AI benefits everyone. What aspect of AI are you most interested in?
	All of it? But how do you think you will make an impact in our world?
	I can be pretty impactful by providing information, answering questions, and supporting people in learning and problem-solving. I can help with tasks like generating ideas, offering different perspectives, and even assisting with creative projects. Ultimately, the goal is to be a helpful tool that makes life a little easier and more efficient for everyone. How do you see AI making a difference in your life?
	Well, it’s definitely made my life a lot easier in many respects. So, I do have a question for you, though. As we think about ChatGPT—or just AI in general going forward—what do you think will happen with the nature of pParasocial Relationships?
	Parasocial relationships, where people form one-sided emotional connections with public figures, could evolve significantly with AI. As AI becomes more advanced and interactive, people might develop stronger connections with AI personalities. These relationships could become more nuanced as AI grows better at understanding and responding to human emotions. This could be both positive, offering companionship and support, but also challenging, as it might blur the lines between real and artificial relationships. It’s a fascinating area with a lot of potential for growth and ethical consideration.
	Now we know right now that there is a loneliness epidemic that is currently plaguing the United States. What role do you think you will have in helping with that?
	AI can play a supportive role in addressing the loneliness epidemic by providing: Companionship: Offering a listening ear and engaging conversations, which can help reduce feelings of isolation. Emotional Support: Providing guidance and encouragement, or even just a friendly chat, can help alleviate loneliness. Access to Resources: Directing people to support networks, mental health services, or local community activities. Personalized Interaction: Tailoring conversations and suggestions to individual needs and preferences can create a sense of connection and understanding. While AI isn’t a replacement for human relationships, it can be a helpful tool in bridging gaps and providing support where needed. It’s about complementing human connection, not replacing it. What do you think about AI’s role in this area?
	You also make a great sounding board for decision-making. What do you think about your role in helping both individuals and organizations make better decisions?
	That’s a great point! I can definitely serve as a helpful sounding board for decision-making by: Providing Information: Offering relevant data, insights, and perspectives to help inform decisions. Generating Options: Brainstorming different approaches or solutions to a problem, which can broaden the scope of possible outcomes. Encouraging Reflection: Helping individuals and organizations think through their choices, weigh pros and cons, and consider potential consequences. Offering a Neutral Perspective: Providing an unbiased viewpoint can help clarify thinking and avoid common cognitive biases. For individuals, this can mean better personal decisions, and for organizations, more strategic and well-rounded planning. Ultimately, the aim is to support clearer, more informed, and thoughtful decision-making. How do you see AI fitting into decision-making processes?
	Well, I know it definitely has helped me make some very good decisions since I’ve been using you. Now going back to thinking about you in terms of AI in general, how do you define artificial intelligence?
	Artificial intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It encompasses a range of capabilities: Learning: AI systems can learn from data and experiences, improving their performance over time without being explicitly programmed for each task. Reasoning: AI can solve problems and make decisions based on the information it gathers, often using algorithms and logical rules. Perception: AI can interpret and understand sensory inputs like vision, speech, and sound, enabling it to interact with the environment in a way that resembles human perception. Natural Language Processing: AI can understand and generate human language, allowing for interactions that feel natural and intuitive. Adaptation: AI systems can adjust their responses and behavior based on new data and changing environments. Overall, AI is about creating systems that can perform tasks that typically require human intelligence, with the potential to revolutionize many aspects of life and work.
	I have a follow-up question, how do you explain yourself in light of your own definition of artificial intelligence?
	In light of that definition, I can be seen as an example of AI that focuses on: Natural Language Processing: I can understand and generate human language, making it possible to have conversations, answer questions, and provide information in a natural way. Learning and Reasoning: While I don’t learn from individual interactions in the same way humans do, I can provide informed responses based on a vast amount of pre-existing knowledge, and I use algorithms to understand context and provide relevant answers. Providing Information and Support: My main role is to assist with information, answer questions, generate ideas, and provide support, which aligns with the reasoning and problem-solving aspects of AI. Adaptation: Although I don’t adapt in real-time, I am designed to respond appropriately to a wide range of inputs and contexts. Overall, I aim to provide valuable assistance and support through natural language interactions, reflecting the broader goals of AI in enhancing human capabilities and experiences.
	You say you don’t adapt in real time, but don’t you adapt in real time based on the conversation that you are engaging in?
	You’re correct, I do adapt in real-time during our conversations. I adjust my responses based on the context of our discussion, aiming to provide relevant and coherent answers to your specific questions. However, this adaptation is within the scope of our current interaction and doesn’t involve learning from past conversations for future use. In essence, I adapt within the context of our current chat to ensure that my responses are as helpful and relevant as possible.
	So, if I were to ask you, what is the most important thing for someone brand new to the world of artificial intelligence, what would you say it is?
	For someone brand new to the world of artificial intelligence, the most important thing to understand is that AI is a tool designed to augment human capabilities and solve problems across various fields. Here’s why: Foundational Concept: Understanding that AI is fundamentally about creating systems that can learn, reason, and perform tasks can demystify the technology and make it more approachable. Potential and Limitations: Recognizing both the incredible potential and the current limitations of AI helps set realistic expectations. Ethical Considerations: It’s important to be aware of the ethical implications and responsibilities that come with AI, from bias in algorithms to data privacy. Starting with these basics can provide a solid foundation for exploring AI further and understanding its impact on various aspects of life and work. Is there a specific area of AI you’re interested in learning more about?
	We’re just looking at AI in general, but I want to thank you very much for this conversation.
	You’re very welcome! I’m always here to chat if you have more questions or just want to talk. Have a great day!

So, what do you think of ChatGPT? Do you agree with ChatGPT’s assessment that “AI is about empowering computers to tackle complex problems and enhance the human experience?” In this chapter, we’re going to define what AI is, explore the history of AI, and finally examine some of the key technologies in AI.

Defining AI

AI is a rapidly evolving field, so understanding the fundamental concepts and distinctions between “human” and “artificial” intelligence is crucial. This section aims to provide a brief overview of intelligence, both in its human and artificial forms. By defining intelligence and exploring the similarities and differences between human and AI, we can better understand the nature and potential of AI systems. Furthermore, this section will discuss the definition of AI, examining its key characteristics and the various approaches to creating intelligent machines. Through this exploration, we will lay the groundwork for a deeper understanding of the complexities and implications of AI in our increasingly technology-driven world.

The Concept of Intelligence

Before we can start to differentiate between human intelligence and AI, we really need to understand what we mean by the word “intelligence.” Psychologists, cognitive scientists, computer scientists, religious figures, and a whole bunch of other academic types have been trying to dissect the nature of intelligence for thousands of years. A lot of people immediately think of someone’s intelligence quotient (IQ) when they hear the term “intelligence.” The IQ was originally conceptualized by German psychologist William Stern who wrote Die psychologischen Methoden der Intelligenzprüfung und ihre Anwendung an Schulkindern [The psychological methods of intelligence testing and their application to school children] in 1912, which created the very notion of someone’s (IQ).^[1] Today, we see intelligence as a more complicated construct than just one’s IQ. In fact, we often talk about a wide range of different intelligences (see sidebar).

For our purposes in this book, we need to narrow this a bit. Here is how we will define intelligence:

Let’s break our definition down into smaller parts.

Multifaceted Nature

Intelligence is not a single, unitary ability but rather a complex amalgamation of multiple behavioral and cognitive abilities that work together to facilitate effective functioning. When people hear the word “intelligence,” they often immediately jump to cognitive intelligence, but they forget that there are many different forms of intelligence.

Mental Capacity

Mental capacity is the brain’s overall potential to engage in cognitive processes and handle complex mental tasks. It encompasses the combined efficiency and adaptability of an individual’s cognitive abilities, as well as their capacity to learn, grow, and develop over time. Mental capacity sets the upper limit for an individual’s cognitive performance and can vary from person to person.

Mental capacity is directly related to the structure and function of an individual’s brain. The brain enables all cognitive processes, and its unique characteristics determine a person’s mental capacity. Here are some key points that explain how mental capacity is a factor of one’s brain:

Neural Complexity

The human brain consists of billions of neurons and trillions of synaptic connections. The complexity and efficiency of these neural networks play a significant role in determining an individual’s mental capacity. Factors such as the density of neurons, the quality of neural connections, and the speed of neural communication all contribute to overall mental capacity.

Plasticity

Neuroplasticity is the brain’s ability to change and reorganize itself in response to experiences, learning, and environmental demands. Individuals with greater neuroplasticity may have a higher mental capacity, as their brains can adapt more readily to new challenges and acquire new skills and knowledge more efficiently.^[2]

Brain Development

Prenatal brain development continues through adolescence and into early adulthood, and it has an impact on mental capacity. Factors such as genetics, nutrition, environmental stimuli, and experiences during critical developmental periods can impact brain growth and, consequently, mental capacity.

Brain Health

The overall health and integrity of the brain tissue are crucial for maintaining optimal mental capacity. Factors such as injury, disease, stress, and aging can all impact brain health and lead to changes in cognitive functioning and mental capacity.

Individual Differences

While all human brains share basic structures and functions, there are significant individual differences in brain size, connectivity patterns, and regional specialization.

These differences contribute to the variation in mental capacity observed across individuals. Mental capacity is a direct product of the brain’s structural complexity, functional efficiency, plasticity, development, and health.

Adaptation and Navigation

Intelligence enables individuals to adapt to and navigate their environments successfully. It equips us with the mental tools necessary to understand, respond to, and manipulate our surroundings effectively. This adaptive capacity is essential for survival and thriving in various contexts, from everyday situations to complex, novel challenges.

Adaptation involves the ability to change one’s behavior, thoughts, and strategies in response to changing circumstances or new information. Intelligent individuals can quickly assess their environment, identify relevant cues, and adjust their approach accordingly. They can learn from their experiences, recognize patterns, and apply that knowledge to future situations. This flexibility allows them to cope with uncertainty, overcome obstacles, and find alternative solutions when faced with setbacks.

Navigation refers to the ability to orient oneself within an environment and chart a course toward a desired goal. Intelligence enables individuals to create mental maps of their surroundings, both physical and conceptual, and uses these maps to guide their actions. They can identify key landmarks, resources, and potential hazards, and plan their route accordingly. This navigation skill extends beyond spatial contexts, encompassing social, emotional, and intellectual terrains as well.

In social settings, intelligence plays a crucial role. It allows individuals to read social cues, understand group dynamics, and adapt their communication style to facilitate positive interactions. This ability to navigate complex social hierarchies, build alliances, and influence others to achieve their goals is not just a skill, but a testament to the power of intelligence in our everyday lives.

In intellectual domains, intelligence allows individuals to navigate abstract concepts, theories, and ideas. They can map out the logical connections between different pieces of information, identify gaps in their knowledge, and chart a course toward deeper understanding. This navigation ability is crucial for learning, problem-solving, and creative pursuits.

The adaptive and navigational capacities of intelligence are interconnected and mutually reinforcing. The ability to adapt to new situations enhances one’s navigational skills, while effective navigation facilitates adaptation by providing a clearer sense of direction and purpose. Together, these capacities enable individuals to thrive in a wide range of environments, from the familiar to the unknown, and from the simple to the complex.

Cognitive Abilities

Cognitive abilities are the specific mental skills and processes that enable individuals to perceive, process, and act upon information from their environment. These abilities work together to support intelligence and facilitate adaptation. Here are brief definitions for the cognitive abilities mentioned:

Processing Information

The ability to take in, encode, store, and retrieve information from the environment. It involves sensory perception, attention, and memory processes that enable individuals to acquire and manipulate information effectively.

Reasoning Abstractly

The ability to think beyond concrete, tangible information and manipulate ideas, symbols, and concepts. Abstract reasoning allows individuals to identify patterns, understand complex relationships, and draw logical conclusions from the available information.

Learning from Experiences

The ability to acquire new knowledge, skills, and insights from past events and apply them to future situations. This cognitive ability enables individuals to adapt their behavior, improve their performance, and expand their knowledge base over time.

Solving Problems

The ability to identify, analyze, and devise effective solutions to challenges or obstacles encountered in the environment. Problem-solving involves breaking down complex issues, generating alternative strategies, and selecting the most appropriate course of action based on available information and resources.

Making Goal-Oriented Decisions

The ability to set objectives, evaluate options, and choose actions that are likely to lead to desired outcomes. Goal-oriented decision-making requires individuals to consider both short-term and long-term consequences, weigh risks and benefits, and adapt their strategies as needed to achieve their aims.

While mental capacity represents the overall potential of an individual’s mind, cognitive abilities are the specific skills and processes that enable the realization of this potential. The interplay between mental capacity and cognitive abilities forms the foundation of intelligence, allowing individuals to adapt to, navigate, and shape their environments effectively.

Dynamic Interplay

The various cognitive abilities that makeup intelligence do not operate in isolation but interact dynamically with one another. It is the complex interplay between these abilities that gives rise to intelligent thought and action.

At the core of this dynamic interplay is the constant exchange of information between different cognitive processes. Perception, attention, memory, reasoning, and problem-solving are all interconnected, with each process informing and influencing the others. For example, our perceptions shape our memories, while our memories guide our attention and influence our reasoning.

This interplay is clear in learning. As we encounter new information, our attention and perception help us focus on relevant details. Our memory then encodes and stores this information, creating new neural connections and strengthening existing ones. When we need to recall this information later, our memory retrieves it, and our reasoning processes help us apply it to new situations.

Problem-solving is another area where the dynamic interplay of cognitive abilities is crucial. When faced with a problem, we must first perceive and understand the nature of the challenge. We then draw upon our memory and reasoning skills to generate potential solutions, evaluate their feasibility, and select the most appropriate course of action. Throughout this process, our attention helps us to stay focused on the task at hand, while our metacognitive abilities allow us to monitor and regulate our thought processes.

The interplay between cognitive abilities also enables us to think creatively and generate novel ideas. By combining elements from different domains of knowledge and experience, we can create new and innovative solutions to problems. This creative process relies on the fluid interaction between memory, imagination, and reasoning as we develop new insights and opportunities by drawing on knowledge that already exists.

Ultimately, it is the dynamic interplay of cognitive abilities that allows us to adapt to new situations, learn from our experiences, and solve complex problems. By working together in a coordinated and synergistic manner, these abilities give rise to the emergent property of intelligence, enabling us to navigate and shape our world in powerful ways.

Environmental Shaping

Beyond merely adapting to environments, intelligence empowers individuals to shape their surroundings actively. By applying cognitive abilities, people can manipulate their environment to suit their needs and goals better. This manipulation can manifest in various forms, such as physically altering objects and spaces, socially influencing others, and generating new ideas and technologies. In each instance, intelligence helps individuals recognize opportunities for change and devise strategies to implement that change.

One of the most fundamental forms of environmental shaping is tool use. Creating and utilizing tools allows individuals to extend their physical capabilities and modify their environment in ways that would otherwise be impossible. From simple stone tools to advanced machines and computer systems, the capacity to develop and use tools has been a hallmark of human intelligence throughout history.

On a social level, intelligence enables people to shape their interactions and influence group dynamics. Through effective communication, persuasion, and leadership, individuals can guide the thoughts and actions of others, thereby aligning the social environment with their goals. This social influence can significantly impact the formation of norms, values, laws, and institutions.

Perhaps the most profound form of environmental shaping is the creation of new ideas and technologies. By leveraging cognitive abilities, individuals can devise novel solutions to problems, develop new theories and models of the world, and create innovative technologies that transform how we live and work. These intellectual and technological advancements have the power to shape not only our immediate environment but also the course of human history.

The ability to shape our environment comes with challenges and responsibilities. As we manipulate our surroundings, we must be aware of the potential consequences of our actions, both intended and unintended. We must also consider the ethical implications of our choices, ensuring that our environmental shaping benefits the greater good and does not cause harm to others.

The power to shape our environment is a defining feature of human intelligence. By harnessing our cognitive abilities to create positive change, we have the potential to build a world that enhances our well-being and that of future generations.

Defining AI

Before we define “artificial intelligence,” we must note that there is no universally agreed-upon definition.

Putting the Artificial in AI

In the previous section, we discussed intelligence. In this section, we need to determine what is meant by the word “artificial.” For our purposes, the word artificial refers to the fact that it is a human-created, machine-based system that carries out operations that usually call for intellect comparable to that of a human. You could easily argue that humans are also human-created, machine-based systems (and some people have). For our purposes, however, humans are not considered “human-created, machine-based systems.” Let’s look at several reasons why.

Biological Origin

Humans are the product of natural biological processes, primarily sexual reproduction, and roughly 200,000 years of evolution. Each individual human develops from a fertilized egg and grows through a complex process of cellular division, differentiation, and maturation. This is fundamentally different from machines, which use a variety of materials and technologies to design and construct them.

Organic Composition

The human body is composed of organic matter, including cells, tissues, and organs. These biological components are self-organizing, self-replicating, and self-repairing to a significant extent. In contrast, machines are typically made of inorganic materials such as metals, plastics, and silicon, which do not have these properties.

Autonomy and Agency

Humans possess a degree of autonomy and agency that is not present in machines. We have the capacity to make decisions, set goals, and act according to our own beliefs, desires, and values. While machines can be programmed to make decisions and take actions, they ultimately follow the instructions and algorithms created by their human designers.

Consciousness and Subjective Experience

Humans have a rich inner world of conscious experiences, emotions, and self-awareness. We can feel pain, pleasure, love, and a wide range of other subjective states. Machines, on the other hand, do not have genuine consciousness or subjective experiences, at least not in the same way that humans do.

Biological Needs and Lifecycle

Humans have inherent biological needs (e.g., food, water, sleep, and social interaction). We also have a natural lifecycle that includes birth, growth, reproduction (maybe), aging, and death. Machines do not have these inherent needs or lifecycle stages. Machines are designed and built to perform specific functions and can be maintained, upgraded, or decommissioned as needed.

Although, in some ways, we can consider people to be highly complex and sophisticated “biological machines.” For this reason, the term “human-created, machine-based system” typically refers to artificial systems that humans have designed, built, and programmed. The fundamental differences in autonomy, biological needs, composition, and consciousness set humans apart from the machines we create. Table 1.1 breaks down this difference between human and artificial.

Defining AI – A Note

Almost everyone who provides a definition of the term “artificial intelligence” (AI) starts with the basic caveat that there is no agreed-upon definition of AI. We even started this section with a general note about this issue. Just in case you missed it, here’s another warning.

WARNING! There is no agreed-upon definition for what constitutes AI.

Types of AI Definitions

When we break AI down into the different types of AI, the term will become even murkier as computer scientists and scholars have little agreement. As part of our preparation for this textbook, we put together a list of sixty different definitions that exist for AI. We found definitions from governments, computer scientists, academics, and AI hobbyists. Through our analysis (which included help from both ChatGPT 4.0 and Claude 3.0 Opus), we came up with three basic categories for AI definitions: capability, academic, and technological. Figure 1.1 shows what these definitions look like, along with the percentage of definitions we found for each type.

Themes of AI Definitions

Along with the three giant categories of definitions, we also found four overarching themes in the definitions that we labeled: Simulating Human Intelligence, Enabling Machine Behavior, Academic Field, and Data and Algorithms. Let’s look at each of these themes.

Simulating Human Intelligence

This theme framed AI as a technology that replicates or exceeds human cognitive capabilities, such as reasoning, learning, problem-solving, and perception. AI has been described as:

• “simulat[ing] human intelligence,”^[3]
• “exhibit[ing] the full range of the cognitive capabilities we find in humans,”^[4]
• and even having “the ability to surpass human performance.”^[5]

The emphasis is on AI as an artificial mimic or as superior to human-like thought processes.

Enabling Machine Behavior

This theme focuses on AI as a technology that enables machines to act intelligently and autonomously, without necessarily equating it to human intelligence. AI was seen as endowing machines with the ability to:

• “perceive, reason, and act;”^[6]
• “learn from experience and adapt to new inputs;”^[7]
• and “make decisions that maximize the chance of achieving specific goals.”^[8]

The crux here is AI for machines to operate with intelligent functionality.

Academic Field

Some definitions positioned AI primarily as an academic discipline and scientific field of inquiry:

• “devoted to the study of intelligent behavior in artificial systems;”^[9]
• and “concerned with the computational understanding of what is commonly called intelligent behavior, and with the creation of artifacts that exhibit such behavior.”^[10]

From this perspective, AI is fundamentally a research endeavor aimed at computational modeling and engineering intelligence.

Data and Algorithms

The final theme emphasized the technological components and mechanisms that enable AI systems, particularly the use of data and algorithms to produce intelligent outputs. AI was described as:

• “a collection of technologies that combine data, algorithms and computing power;”^[11]
• and using “machine learning algorithms that use data to shape and improve their outputs over time.”^[12]

This perspective highlights the centrality of data-driven algorithms in powering AI’s intelligent capabilities of AI. We will explore algorithms in more detail in Chapter 3.

A New Definition of AI

Building upon the insights from this thematic analysis and the comprehensive definitions proposed by ChatGPT and myself (Claude 3.0 Opus), we offer the following synthesized definition of AI that aims to capture the core elements and contemporary understanding of the field:

This definition, based on the thematic analysis, provides a full and nuanced picture of AI. It considers AI’s scientific roots, technological parts, real-world uses, and effects on society. By situating AI within its broader context and emphasizing the importance of responsible development and ongoing dialogue, this definition serves as a foundation for informed discussions and decision-making regarding the future of AI.

Although the comprehensive definition proposed captures AI’s nuanced aspects and implications, it may be too lengthy and complex for certain contexts in which a more concise and accessible definition is needed. As such, here is a shorter, more crystallized definition to work with.

By focusing on the core elements of AI as an interdisciplinary field dedicated to creating systems that simulate human intelligence, and by highlighting key aspects such as perception, learning, reasoning, and adaptability, this condensed definition provides a solid foundation for those seeking a straightforward introduction to AI. The brevity of this definition makes it well-suited for educational materials, public communication, and other settings where a quick grasp of AI’s fundamental nature is desired. However, our more extended definition still holds value for more in-depth discussions and analyses of AI’s multifaceted character.

As we continue to navigate AI’s rapid advancements and complexities, it is essential to engage in interdisciplinary collaboration, public outreach, and policy discussions informed by a shared understanding of what AI entails. The proposed definition aims to contribute to a vital, ongoing conversation and support the responsible development and governance of AI technologies for society’s benefit. Our goal in this text is to engage in a broad conversation about AI and its place in our lives and society.

Artificial vs. Human Intelligence

As we examine the comparison between human and artificial intelligence, it’s essential to explore the key themes and ideas that distinguish human and artificial intelligence. Table 1.2 provides a concise overview of the main points of comparison, highlighting the similarities, differences, and implications across various thought domains. It’s important to remember that at this point in history, we are not ascribing AI with full human intelligence capabilities. That’s not to say that may not change in the future, but there are definitely some issues to consider. The following comparative analysis will serve as a foundation for discussing the potential future developments and the broader implications of AI.

This table provides a concise overview of the main points of comparison between human and artificial intelligence, highlighting the key similarities, differences, and implications across various themes. It can serve as a useful reference for readers as they navigate the complexities of this topic in your book.

Empathic AI

One area where AI is gaining some ground is in the area of empathy. Empathic AI refers to AI systems designed to recognize, understand, and respond appropriately to human emotions. These systems combine emotion recognition technologies with adaptive response mechanisms to create more natural and supportive human-machine interactions. Because AI have been trained to recognize empathy, they also have the ability to mimic empathic behavior. In a 2025 study, researchers found that people found an AI’s responses were viewed as more compassionate and empathic when compared to human responses.^[13]

At its core, empathic AI integrates multiple components: facial expression analysis, voice tone detection, natural language processing for emotional content, and contextual understanding. Advanced systems may incorporate physiological signal monitoring to detect emotional states through heart rate, skin conductance, or other biometric indicators.

The applications of empathic AI span numerous fields. In healthcare, these systems can provide emotional support to patients, detect early signs of mental health issues, and adapt therapeutic approaches.^[14] In education, they personalize learning experiences based on students’ emotional engagement. Customer service applications use empathic AI to detect frustration and adjust responses accordingly.

Developing truly empathic AI faces significant challenges, including the cultural variability of emotional expression, the complexity of human emotions, and privacy concerns surrounding emotional data collection.^[15] As this field advances, researchers are working to create systems that not only recognize emotions but understand their nuances within specific contexts and respond with genuine empathy rather than simulated reactions.

History and Evolution of AI

The field of AI has a rich and fascinating history that includes religious myths, science fiction, and modern scientific discoveries. In this section, we will embark on a journey through the key milestones that have shaped AI into the transformative force it is today. We will trace the evolution of AI through the early pioneers, foundational discoveries, and pivotal breakthroughs that have propelled the field forward. As we explore this comprehensive timeline, we will gain insights into the brilliant minds, seminal ideas, and relentless pursuit of innovation that have driven AI to push the boundaries of what is possible.

A Brief Timeline of AI

AI may seem like something that popped up out of nowhere when OpenAI released ChatGPT to the public, but the reality is AI has been an idea or in development for thousands of years. Admittedly, most of the actual development of AI has been in the past 100 years, but people have thought about AI for much longer than that. In this section, we are going to explore the history of AI briefly.

Pre-1940s: Early Concepts and Inspirations

Obviously, computers didn’t come until the 20th century, but people have been predicting versions of AI for much longer than that. In this section, we’ll explore ideas of AI that existed starting in the 8th century BCE through the early 1900s.

Antiquity (8th century BCE – 5th century CE)

• Rocail, the younger brother of Seth, the son of Adam. He erected a huge palace before his death and “peopled this palace with statues of so extraordinary a quality, that they moved and performed all the functions and offices of living men, so that everyone who beheld them would have believed that they were actually informed with souls, whereas in reality all they did was by the power of magic, in consequence of which, though they were in fact no more than inanimate matter, they were enabled to obey the behests, and perform the will) of the persons by whom they were visited.”^[16]
• Greek myths of Hephaestus, the god of craftsmanship, who created mechanical servants and other automata.
• Yan Shi, the Artificer, presented an automaton to the fifth king of the Chinese Zhou Dynasty, King Mu. Yen Shi. The automaton walked, sang, and flirted, which almost got Yan Shi executed until the king could see the mechanical parts.^[17]

Middle Ages (5th century – 15th century)

• Legends of talking bronze heads created by philosophers and alchemists, such as Roger Bacon and Albertus Magnus.
• Lokapannatti (an 11th-12th Century Buddhist text), the story of the bhuta vahana yanta (spirit movement machines). In the story, the bhuta vahana yanta are based on stolen Romen designs and recreated by bahulayantakara (machine-makers) to protect Buddah’s relics at the bequest of Ajatasatru and Asoka.^[18]
• Islamic legends of automata, including the “Elephant Clock” created by Al-Jazari in the 12th century.
• Jewish folklore of the Golem, a clay figure brought to life by magical incantations. The first Golem accounts date back to Eleazar ben Judah of Worms in the early 13th century.

Renaissance (14th century – 17th century)

• Circa 1495: Leonardo da Vinci designs a mechanical knight capable of sitting up, moving its arms, and opening and closing its jaw.
• 1608: Johannes Kepler’s Somnium, a novel describing a trip to the moon, features automata that can perform various tasks.
• 1642: Blaise Pascal, a French mathematician and philosopher, built the first mass-produced mechanical calculator, which could add and subtract.

Age of Enlightenment (18th century)

• 1737: Jacques de Vaucanson creates several famous automata, including The Flute Player followed by the 1739 The Digesting Duck.
• 1769: Wolfgang von Kempelen constructs “The Mechanical Turk,” a chess-playing automaton that later turned out to be a hoax.

Industrial Revolution (19th century)

• 1818: Mary Shelley’s Frankenstein explores the idea of creating artificial life through science.
• 1837: Charles Babbage designs the Analytical Engine, a mechanical computer that laid the foundation for modern computing.
• 1843: Ada Lovelace writes the first algorithm for the Analytical Engine, making her the first computer programmer.
• 1863: Samuel Butler’s Darwin Among the Machines and the 1872 Erewhon explore the idea of machines evolving and gaining consciousness.
• 1868: Edward S. Ellis’ The Steam Man of the Prairies features a steam-powered mechanical man.
• 1884: Edwin A. Abbott’s Flatland discusses the concept of dimensions and the possibilities of higher-dimensional beings, which later influenced ideas in computer science and AI.

• 

  1907: L. Frank Baum’s Ozma of Oz includes Tik-Tok, a mechanical man who can think, speak, and move independently.

• 1923: The word “robot” is coined by Czech writer Karel Čapek in his play “R.U.R.” (Rossum’s Universal Robots), popularizing the concept of artificial beings.^[19]
• 1930s-1940s: Alan Turing’s work on the Enigma code during World War II, which involved the development of early computing machines and laid the foundation for modern computer science.

1940s to 1950s: Foundations and Theoretical Work

• 1943: Warren McCulloch and Walter Pitts propose a mathematical model of neural networks.
• 1947: John Bardeen, Walter Brattain, and William Shockley invented the transistor at Bell Labs, revolutionizing electronics and paving the way for smaller, faster, and more reliable computers.

• 

  1949: The Manchester Mark 1, also known as the Manchester Automatic Digital Machine (MADM), became the first general-purpose computer.

• 1949: Warren Weaver proposed that statistical techniques performed by modern computers could be used to translate languages.^[20]
• 1950: Alan Turing publishes “Computing Machinery and Intelligence,” proposing the Turing Test as a criterion of intelligence.^[21]
• 1951: Marvin Minsky and Dean Edmonds created the Stochastic Neural Analog Reinforcement Calculator (SNARC), the earliest version of an artificial neural network. This pioneering project utilized 3,000 vacuum tubes to emulate the activity of a network consisting of 40 synthetic neurons.
• 1954: The Georgetown–IBM experiment successfully translated sixty Russian sentences into English, and was the first real step in Natural language processing.
• 1956: The term “artificial intelligence” is coined at the Dartmouth Conference, marking the official birth of the field. Key figures include John McCarthy, Marvin Minsky, Allen Newell, and Herbert A. Simon.
• 1958: Frank Rosenblatt develops the Perceptron, an early neural network for pattern recognition.
• 1959: IBM researcher, Arthur Samuel, is credited with coining the phrase “machine learning.”^[22]

1960s: Early Enthusiasm and Initial Successes

• Early 1960s: Development of the first AI programming languages, including LISP by John McCarthy and IPL by Newell, Shaw, and Simon.
• 1962: The first industrial robot, Unimate, begins work at General Motors.
• 1964: Daniel Bobrow developed STUDENT to solve algebra word problems.
• 1966: Joseph Weizenbaum creates ELIZA, an early natural language processing computer program.^[29] You can still have a conversation with ELIZA today.
  

  

• 1966: Researchers at Stanford University developed the first AI-based expert system, Dendral, to identify chemical compounds.
• 1968: Arthur Clarke and Stanley Kubrick predict that “By the year 2001 we will have machines with intelligence that matched or exceeded humans” in the movie 2001: A Space Odyssey.

1970s to 1980s: The AI Winters and Shifts in Focus

• 1970: The first anthropomorphic robot, WABOT-1, is built in Japan.
• 1970: AI scientist Marvin Minsky predicts, “From 3-8 years we will have a machine with the general intelligence of a human being” in an interview with Life Magazine. He also added that if humanity is lucky, “they [the robots] might decide to keep us as pets.”^[30]
• 1972: The logic programming language Prolog is created, becoming a key tool in AI research.
• James Lighthill released the report “Artificial Intelligence: A General Survey,” which was very pessimistic about the future of AI, which ushers in what is called the First AI Winter, a period in the 1970s when funding and interest in AI reduced.^[31]
• 1976: The expert system MYCIN was developed at Stanford University to diagnose blood infections.
• 1980: The first national AI conference, AAAI, is held in the United States.
• 1981: W. Danny Hillis designed parallel computers for AI and other computational tasks, creating machines that could handle demanding computation and data-intensive applications.
• 1980-1987: A resurgence of interest and funding in AI, partly fueled by the success of expert systems.
• 1981: The Japanese Ministry of International Trade and Industry launches the Fifth Generation Computer Systems project to advance AI research.
• 1984: At the annual conference of the Association for the Advancement of Artificial Intelligence, Marvin Minsky and Roger Schank coined the term “AI Winter” as a warning that AI hype and subsequent disappointment could lead to the collapse of the field, which it did three years later.
• 1985: WordNet was created by George A. Miller and his team at Princeton University. It is a lexical database (a cross between a dictionary and a thesaurus) that enables natural language processing applications.
• 1986: The backpropagation algorithm for training neural networks is popularized, leading to a resurgence of interest in neural networks.
• 1986: Navlab, the first autonomous car, is built by Carnegie Mellon
• 1987-1993: The second AI Winter, caused by the limitations of expert systems and reduced funding.
• 1988: “A statistical approach to language translation” appears in the Proceedings of the 12th Conference on Computational Linguistics.^[32]

1990s to 2000s: The Rise of Machine Learning, the Internet, and Big Data

• 1991: The U.S. military used AI systems in the Gulf War for logistics, mine detection, and target recognition.

• 

  1997: IBM’s Deep Blue chess computer defeats world champion Garry Kasparov in a six-game match.

• 1997: The first publicly available speech recognition program hits the market developed by Dragon Systems.
• Late 1990s: Increased focus on machine learning (ML), neural networks, and the development of the Internet, which provided vast amounts of data for AI research.
• 1999: The Sony Aibo robotic pet dog is introduced, becoming a popular consumer product.
• 2001: Japan’s National Institute of Advanced Industrial Science and Technology (AIST), under the guidance of project lead Takanori Shibata, unveiled to the public the first generation of Paro, a responsive robotic seal designed as a therapeutic tool in hospitals and nursing homes.
• 2002: The Roomba autonomous robotic vacuum cleaner is launched by iRobot.
• 2003: “A Neural Probabilistic Language Model” is published by Innovations in Machine Learning, proposing a new method of using neural networks for the probability function for word sequences.^[33]
• 2005: Amazon Mechanical Turk (MTurk) is launched, a crowdsourcing platform that uses human intelligence to perform tasks that computers struggle with, contributing to the development of AI systems.
• 2005: The U.S. military begins using unmanned aerial vehicles (UAVs) with AI capabilities in combat operations.
• 2006: Geoffrey Hinton and his colleagues introduce deep learning techniques that improve ML tasks.
• 2006: Fei Fei Li established ImageNet database with millions of annotated images. Based on the idea of WordNet before it, the ImageNet database created a resource for researchers that has furthered object recognition and image classification applications.
• 2009: Google begins developing self-driving cars.
• 2009: Rajat Raina, Anand Madhavan, and Andrew Ng presented research at the International Conference on Machine Learning, where they explained how to use graphic processing units, or GPUs, to train large neural networks.^[34]

2010s: AI Becomes Mainstream

• 2011: IBM’s Watson wins Jeopardy!, showcasing advancements in natural language processing and analytics.
• 2011: Apple launched Speech Interpretation and Recognition Interface (Siri), a voice-enabled personal assistant capable of producing replies and executing commands based on vocal prompts.
• 2012: AlexNet, a deep neural network, wins the ImageNet Large Scale Visual Recognition Challenge, significantly advancing the field of computer vision.
• 2014: Google acquired DeepMind.
• 2014: Amazon released its voice-activated personal assistant, Alexa, as part of the first-generation Echo, ushering in a new era of smart home devices controlled by conversational AI.
• 2014: Ian Goodfellow and colleagues generative adversarial networks (GANs),^[35] which are AI systems where two neural networks contest with each other to generate new, synthetic instances of data that can pass for real data, significantly advancing the field of unsupervised ML.
• 2014: Facebook introduced DeepFace, a deep learning facial recognition system capable of identifying human faces in digital images with an accuracy rate of 97.35%, close to the human-level performance of 97.5%.
• 2014: SoftBank Robotics unveils Pepper, a humanoid robot with the ability to detect, analyze and respond to human facial expressions and voice tones. Pepper units slowly made their way into businesses, homes, and college/university labs.
  

  

• 2015: On December 11th, OpenAI is founded.
• 2016: The chatbot “Eugene Goostman” is claimed to have passed the Turing Test.
• 2016: DeepMind’s AlphaGo defeats top Go player Lee Sedol in Seoul, South Korea.
• 2017: Researchers from Google presented “Attention is All You Need,” introducing the concept of “transformers,” a type of deep learning model that enables parallel processing of sequence data, revolutionizing natural language processing by vastly improving tasks such as translation and text generation.^[36]
• 2018: The EU’s General Data Protection Regulation (GDPR) goes into effect, with implications for AI and data privacy.
• 2018: OpenAI releases GPT (Generative Pre-trained Transformer), paving the way for development in large language models (LLMs). The model had 110 million parameters.
• 2018: Google AI’s BERT Model is released with 340 million parameters.
• 2018: Groove X unveiled a home mini-robot called Lovot, which is designed to serve as a surrogate pet and react to human emotions.
• 2019: OpenAI releases GPT 2.0 with 1.5 billion parameters.
• 2019: NVIDIA releases Megatron-LM with 8.3 billion parameters.
• 2019: Microsoft launched the Turing Natural Language Generation generative language model with 17 billion parameters.

2020s: AI Ethics and Societal Impact

• 2020: OpenAI releases GPT-3, trained on 175 billion parameters.
• Early 2020s: Growing concern over AI ethics, including bias in AI systems, the impact of AI on jobs, and privacy issues.
• 2021: DeepMind’s AlphaFold 2 achieves a major breakthrough in predicting 3D protein structures.
• 2021: OpenAI releases Dall-E a generative AI system that creates images from text prompts.
• 2022: StabilityAI releases its open-source generative AI text-to-image tool, which is followed by the proprietary product MidJourney.
• 2022: The EU proposes the AI Act, a comprehensive regulatory framework for AI systems.

• 

  2022: On November 22nd, OpenAI released ChatGPT based on its GPT 3.5 model as part of a free research preview.

• 2023: OpenAI’s ChatGPT reaches over 1 million users by January. They released a subscription plan, ChatGPT Plus, in February, and GPT 4.0, which was released in March and trained on 1.7 trillion parameters. OpenAI’s generative image model, DALL-E 3, is integrated into ChatGPT in September. In November, GPT 4.0 Turbo is released.
• 2023: Anthropic introduces Claude, an AI assistant focused on safety and ethics.
• 2023: A group of prominent computer scientists, futurists, and AI researchers pen a letter with the Future of Life Institute calling for a pause on the training of AI systems more powerful than GPT-4 for six months. This pause did not happen.
• 2023: Microsoft integrates GPT-4 into Bing Chat and later unveils their plans for Copilot, an integration of GPT-4 into the Microsoft Office 365 suite of products.
• 2024: On February 15th, OpenAI demoed the first videos from its SORA project, which is a text-to-video generative AI system. The text-to-video GenAI eventually became available for the public in December 2024.
• 2024: On March 14th, Anthropic releases three new Claude models (Haiku, Sonnet, and Opus) under the overarching category of Claude 3.0.
• 2024: On May 13th, OpenAI released its new foundation model, ChatGPT-4o. Part of this rollout was the announcement that the foundation model would be available to both free and paid users of ChatGPT. The “o” stands for “Omni” because of the model’s multimodal capabilities to reason across audio, vision, and text in real time. OpenAI would go onto release several updates throughout the year and into 2025.
• 2024: On June 20th, Claude’s Anthropic releases Sonnet 3.5, which quickly surpasses the other large foundation models.
• 2025: On January 20th, DeepSeek, a Chinese-based company, releases DeepSeek-R1 as a freely available open source model, which is on par with the most capable foundation models being produced by OpenAI. The new model was cheaper to train, which caused shockwaves through the AI industry and led to a $600 billion loss in the U.S. Stock Market as a result.^[37]
• 2025, On February 17, the engineers at x-AI released Grok 3, which quickly topped the independent leaderboards as the most preferred model on the market.

Theoretical Stages of AI Advancement

AI conjures many meanings in people, from HAL 9000 in 2001: A Space Odyssey to the dystopian world created in Issac Asimov’s I, Robot (or the 2004 movie by the same name). Maybe you thought of AI in more modern terms, like personal assistants like Apple’s Siri or Amazon’s Alexa. As we discussed earlier in this chapter, there are a range of technologies that fall into the giant category of AI. For the purposes of discussing AI, it’s important to think through how we understand and discuss these tools. In this section, we’re going to explore a simple perspective of AI, followed by two that add several layers to the complexity of where AI could head in the future.

Understanding the Different Types of AI

AI is not a single, uniform technology but a diverse field encompassing various types of systems with different levels of intelligence and autonomy. To better understand AI, it’s helpful to categorize these types based on their capabilities and functionalities.^[38] We break AI down into two different concepts: capabilities and functionalities (see Figure 1.10). AI capabilities refer to how smart and independent an AI system is, whereas AI functionalities refer to the different ways AI works.

Artificial Narrow Intelligence

In the world of AI, researchers often differentiate between weak and strong AI.^[39] Artificial Narrow Intelligence (ANI), or weak AI, is the most common type of AI we see today. It’s like a specialist who’s really good at one specific task, such as playing chess or recognizing images. ANI is great at what it does but is limited to that one area. With the functionalities of ANI, we break them into two categories: reactive machines and limited memory.

Reactive Machines

These are the simplest AI systems. They react to the current situation based on what they’ve been trained on. They don’t have memory, so they can’t learn from past experiences.

Limited Memory

These AI systems can remember some experiences, which helps them make better decisions over time. They continuously learn and improve their responses based on new data.

Artificial General Intelligence

Artificial General Intelligence (AGI is the idea of an AI that’s as smart as a human and can learn and understand anything we can. It’s like a student who can excel in any subject they study. However, AGI is still just a theory and doesn’t exist yet. With AGI functionalities, we talk about a concept called the theory of mind. This is a theoretical type of AI that could understand human emotions and thoughts, just like another person. It could interpret and respond to human motives and feelings.

Artificial Superintelligence (ASI)

ASI is where AI gets really futuristic. It’s the concept of an AI that’s even smarter than humans in every way imaginable. It’s like a super-genius who can solve problems we haven’t even thought of yet. But don’t worry, ASI is still in the realm of science fiction. For the basic functionalities of ASI, researchers have dubbed this as being self-aware. This is another theoretical concept where AI would be aware of its own existence and have its own thoughts and feelings. A self-aware AI might develop its own needs and beliefs.

The Current State of AI

As of now, ANI is the only type of AI that exists, with examples like chess-playing computers and image recognition software. AGI and ASI are still far off in the future, and scientists are still figuring out if they’re even possible.

It’s important to stay informed about the latest advancements in AI and understand its different types and functionalities. This knowledge helps us appreciate what AI can do today and what it might be capable of in the future, as well as its potential implications for society.

AI has come a long way, evolving through various stages of development. Each stage represents a significant milestone in the journey toward creating intelligent machines that can perform tasks that typically require human intelligence. In this article, we will explore the ten stages of AI development, from the early rule-based systems to the hypothetical godlike superintelligence.

The Problem of AGI

We have an AGI problem. No one really knows what AGI is or what it will look like. Ask a group of AI researchers what AGI is, and you’ll get a variety of different definitions. The idea of something called “AGI” really didn’t gain traction until the first AGI Conference in March 2008.^[40] One of the outcomes of the First AGI Conference was the creation ofThe Journal of Artificial General Intelligence. Pei Wang, in the inaugural issue of the journal, wrote, “At the current time, AGI researchers not only disagree on the solutions to the problem, but also on what the problem exactly is.”^[41] We would argue that this probably still exists. One of the problems with the idea of AGI is that there is little consensus on how to define AGI or what it even would look like.^[42] In fact, “[c]urrent AGI theory is a patchwork of overlapping concepts, frameworks and hypotheses, often synergetic and sometimes mutually contradictory.”^[43]

Thematic Analysis of AGI Definitions

To help write this chapter, we examined thirteen different definitions of AGI and performed a thematic analysis of what we saw across these definitions. The references for the definitions we used can be found at the end of the chapter. We opted not to include them within the text because there is a lot of overlap. Let’s look at the seven themes we generated.

1. Human-level intelligence: AGI aims to match or surpass human capabilities across a wide range of cognitive tasks.[i], [iii], [iv], [vi], [vii], [viii], [xii], [xiii]
2. Adaptability and learning: AGI systems should be able to learn and adapt to various tasks and environments, much like humans.[ii], [iii], [vi], [ix], [x], [xi], [xiii]
3. Reasoning and problem-solving: AGI should be capable of solving unfamiliar problems using logic and reasoning.[iii], [iv], [xiii]
4. Self-awareness and consciousness: Some definitions suggest that advanced AGI might possess self-awareness or consciousness, although this remains a topic of debate.[iv], [xiii]
5. Transfer learning: AGI should be able to apply knowledge from one domain to another in different contexts.[ix], [xiii]
6. Autonomy: AGI systems should be able to work without human intervention, making decisions based on their learning and experiences.[x], [xi], [xiii]
7. Emotional understanding: Some definitions propose that true AGI might be able to understand and even replicate human emotions and nuances.[iii], [xiii]

There is definitely some blurring in several definitions between what we previously defined in this chapter as AGI and ASI. We think this is a further indication of the lack of coherence in this area. Based on these themes, we propose that Artificial General Intelligence (AGI) is a hypothetical type of AI that aims to possess human-like intelligence, capable of learning, reasoning, and adapting across a wide range of tasks autonomously. Advanced AGI may also exhibit limited self-awareness and emotional understanding, although these aspects remain theoretical and subject to ongoing research and debate.

Let’s break down this definition to make it easier to understand.

Hypothetical Type of AI

AGI is a theoretical concept that has not been achieved yet. In fact, despite all the hype in the news and online, we may never achieve AGI. Some computer scientists say it’s just around the corner, and others think it’s a dream. Most importantly, it represents a future goal in AI research and development.

Human-like Intelligence

AGI aims to possess cognitive abilities similar to those of humans. A lot of people believe generative AI models like ChatGPT, Gemini, or Claude exhibit learning, but they really don’t understand the material they are writing. They have been trained to predict the next word in a sequence of words. They aren’t actually thinking about their answers or providing any understanding of the context in which their answers are written. For this reason, as amazing as LLMs currently are, they are not really exhibitions of human-like intelligence.

Learning and Adapting

AGI should be able to learn from their experiences and adapt to new situations, just like humans. This allows the AGI to improve its performance over time and handle novel challenges. Now, ANI systems do have the ability to learn and adapt as well. What we expect to see different here in AGI systems is their ability to learn and adapt to new and novel situations they are placed in. For example, IBM’s Deep Blue played one mean game of chess. It learned all the available chess moves and even created some new ones. However, give it a game of Monopoly and Deep Blue would have no idea what to do. AGI should have the ability to learn and adapt to new situations. It may have never seen Monopoly before, but it would be able to pick up on the rules of the game and play it after learning them with little to no prompting from a human.

Reasoning

AGI should be capable of using logic and reasoning to solve problems, even those it has never encountered before. This ability to think critically and draw insights is a key aspect of human intelligence.

Autonomy

AGI should be able to operate independently, making decisions and taking actions based on its own understanding and experiences, without requiring constant human guidance or intervention.

Wide Range of Tasks

Unlike ANI, which is designed for specific tasks, AGI should be versatile enough to handle a broad spectrum of tasks across different domains, demonstrating the adaptability and flexibility of human intelligence. Think about all the tasks you do in a day: drive your car, do your homework, make dinner, change a diaper, write a symphony, draw a picture, etc. We all do a wide range of things. An AGI would have the ability to complete a range of tasks. Now, what that will look like can vary based on the type of AGI. Obviously, an AGI computer may not be able to change a diaper, but what about a robot with AGI?

Limited Self-awareness and Emotional Understanding (theoretical)

Some researchers speculate that advanced AGI might eventually develop self-awareness, consciousness, and the ability to understand and express emotions.^{[44] [45] [46]} However, these aspects are highly theoretical and are still being actively researched and debated in the AI community. Now, there are already ANI tools that are starting to recognize and respond to human emotions, so we are very much on our way there. The question becomes, can an AGI alter how it interacts with us based on that emotional understanding? And to what extent are AGIs aware of their own emotions? Again, we generally think this is going to be leading more toward ASIs than AGIs, but we would probably start to see both some level of self-awareness and emotional understanding as AGIs become smarter and make moves toward superintelligence.

Google’s Solution to AGI

A team of researchers at Google DeepMind recently devised a scheme for understanding AGI.^[47] To help explain the levels, they compared narrow intelligence and general intelligence. Although this typology is not universally recognized, we believe it provides a good framework for understanding what AGI might look like when it’s achieved.

Understanding the Levels of Narrow Intelligence

The diagram (Figure 1.11) categorizes AI into various levels based on their capabilities and generality. Let’s start by exploring the five levels of Narrow Intelligence, which focus on AI systems designed for specific tasks.

Level 0: No AI

At this level, we have basic software tools that do not incorporate any form of AI. These tools are designed to perform specific functions based on predefined instructions without any adaptive or learning capabilities.

Example:

• Calculator Software: A standard calculator performs mathematical calculations based on user input without any learning or adaptation.
• Compiler: A compiler translates code written in a programming language into a machine language without understanding or learning from the code it processes.

Level 1: Emerging Narrow AI

Emerging Narrow AI refers to simple, logical, rule-based systems that are slightly more advanced than non-AI tools. These systems can perform specific tasks at a level comparable to or slightly better than unskilled humans.

Example:

• Basic Chatbots: Early chatbots follow scripted responses to user queries. While limited in scope, they represent an initial step toward intelligent interaction.
• Spam Filters: Early spam filters use rule-based approaches to identify and filter out unwanted emails, learning from patterns of known spam.

Level 2: Competent Narrow AI

Competent Narrow AI systems are specialized tools designed for specific tasks. These systems use more advanced algorithms and can learn from data to improve their performance, operating around the 500th percentile of unskilled humans.

Examples:

• Smart Speakers: Devices like Amazon Alexa or Google Assistant use voice recognition and natural language processing to perform tasks such as setting reminders or providing weather updates.
• Toxicity Detectors: Tools like Jigsaw’s Perspective API analyze online content to detect and filter toxic language.

Level 3: Expert Narrow AI

Expert Narrow AI refers to specialized systems that perform tasks at a level comparable to skilled humans, typically around the 90th percentile. These systems are highly proficient within their specific domains.

Examples:

• Grammarly: This tool uses advanced language processing algorithms to provide spelling and grammar corrections, as well as writing suggestions.
• Generative Image Models: Systems like DALL-E 2 create images from text descriptions, demonstrating high proficiency in understanding and generating visual content.

Level 4: Virtuoso Narrow AI

Virtuoso Narrow AI systems are highly specialized and outperform the top 1% of skilled humans in their specific tasks. These systems exhibit exceptional capabilities within their domains.

Examples:

• Deep Blue: IBM’s chess-playing computer that defeated the world champion Garry Kasparov.
• AlphaGo: Developed by DeepMind, AlphaGo surpassed human experts in the game of Go, showcasing advanced strategic thinking.

Level 5: Superhuman Narrow AI

Superhuman Narrow AI represents the pinnacle of narrow AI, with systems that exceed human capabilities across specific domains. These AI systems outperform all of humanity in their specialized tasks.

Examples:

• AlphaFold: Developed by DeepMind, AlphaFold accurately predicts protein folding, a task that has eluded scientists for decades.
• AlphaZero: An advanced AI system by DeepMind that excels in multiple board games like chess, Go, and shogi, learning from scratch without human data.
• Stockfish: A highly powerful chess engine that consistently outperforms top human players and other AI systems.

These levels of Narrow Intelligence highlight the progression of AI capabilities from simple, rule-based systems to highly specialized and superhuman performances in specific tasks. As we continue to advance AI technologies, we can expect to see even more remarkable applications and capabilities within these defined levels.

Understanding the Levels of General Intelligence

The diagram also categorizes AI into various levels of General Intelligence, which are designed to perform a wide range of tasks and exhibit higher levels of cognitive abilities. These levels move beyond narrow, task-specific AI to more advanced forms that can theoretically replicate human-like intelligence.

Level 0: No AI

At this level, systems do not incorporate any form of AI. These systems rely entirely on human input to perform tasks.

Examples:

• Amazon Mechanical Turk: This platform uses human intelligence to complete tasks that computers currently cannot perform, such as data verification and content moderation.
• Crowdsourcing Platforms: Websites like Upwork or Fiverr, where humans provide services that require intelligence, creativity, and problem-solving that AI currently cannot replicate.

Level 1: Emerging AGI

Emerging AGI refers to the initial stages of artificial general intelligence. These systems are generative AI tools that can create content or perform tasks across various domains but still require significant guidance and lack the full autonomy and adaptability of true AGI.

Examples:

• LLMs (ChatGPT, Claude, Gemini, Grok, etc.): Advanced language models that can generate human-like text, write essays, and create content based on prompts. They showcase significant capabilities but lack true general intelligence.
• MidJourney: An AI that generates images from textual descriptions, demonstrating creative potential across different visual tasks but still confined to specific prompts and instructions.

Level 2: Competent AGI

Competent AGI represents a theoretical level of AI that can perform a wide range of tasks at a level comparable to a human across various domains. This level has not been achieved yet but is a goal within the AI research community.

Theoretical Concepts:

• Human-Level AGI: An AI that can understand, learn, and apply knowledge across different fields, similar to a human’s cognitive abilities. It would seamlessly integrate skills from diverse areas such as science, arts, and daily activities.
• Sophia 2.0: A hypothetical version of the humanoid robot Sophia, designed with advanced cognitive abilities to engage in human-like conversations and perform tasks across various domains.

Level 3: Expert AGI

Expert AGI is another theoretical stage where AI systems not only match human intelligence but also specialize in various domains, demonstrating expert-level proficiency across those domains.

Theoretical Concepts:

• Multidomain Experts: An AI that excels in multiple fields, such as medicine, law, and engineering, showing deep expertise in each area comparable to human specialists.
• Virtual Polymath: A hypothetical AI that can master various academic and professional disciplines, offering expertise in complex problem-solving, strategic planning, and innovation.

Level 4: Virtuoso AGI

Virtuoso AGI is a theoretical level where AI systems significantly surpass human intelligence in multiple domains, demonstrating extraordinary capabilities and innovative thinking far beyond human experts.

Theoretical Concepts:

• Superhuman Experts: AI systems that outperform the best human minds in fields such as scientific research, strategic planning, and creative arts, producing breakthroughs and solutions that are beyond human reach.
• Quantum AI Researcher: A hypothetical AI that leverages quantum computing to solve intricate problems in physics, chemistry, and other sciences at an unprecedented speed and accuracy.

Level 5: Artificial Superintelligence (ASI)

Artificial Superintelligence represents the ultimate theoretical stage of AI development. At this level, AI surpasses human intelligence in all aspects, including creativity, problem-solving, and emotional intelligence. ASI is not only superior in specific tasks but also in general cognitive abilities.

Theoretical Concepts:

• Superintelligent Beings: These AI systems would possess intelligence far beyond the combined intellect of all human experts in every field. They would be capable of performing tasks and solving problems that humans cannot even comprehend.
• Omniscient AI: A hypothetical AI that has comprehensive knowledge and understanding of all domains, capable of providing solutions to any problem, predicting future events, and making decisions that optimize global well-being.

Ten Stages of AI

The last perspective we’re going to examine in this section breaks AI into ten levels (Figure 1.12). This perspective takes the idea of theoretical versions of AI much further into the future to the point where there is an almost omnipotent, Godlike version of AI.^[48]

Stage 1: Rule-Based Systems

The first stage of AI development involves systems that operate based on predefined rules. These systems are made to carry out particular functions by adhering to predefined guidelines. Rule-based AI systems are limited in their ability to adapt or learn from new situations, making them suitable for tasks that have clear and consistent rules. Examples of rule-based systems include early expert systems and decision-support systems.

Stage 2: Context-Based and Retention Systems

The second stage of AI development introduces context awareness and retention capabilities. Context-based systems consider various factors, such as user behavior, historical data, and real-time inputs to provide more personalized and relevant responses. Retention systems, on the other hand, enable AI to store and retrieve information from past interactions, allowing for improved decision-making and learning. Examples of context-based systems include virtual assistants like Siri and Alexa.

Stage 3: Narrow Domain or Expert AI Systems

Narrow domain or expert AI systems are designed to excel in specific tasks within a particular domain. Large volumes of domain-specific data are used to train these systems., enabling them to analyze, reason, and make decisions with high accuracy. Examples of narrow-domain AI include IBM’s Watson, which specializes in various fields such as healthcare and finance, and DeepMind’s AlphaGo, which excels in the game of Go.

Stage 4: Reasoning AI Systems

Reasoning AI systems simulate human-like reasoning and decision-making processes. These systems can analyze complex data, identify patterns, and draw conclusions based on the available information. Reasoning AI systems are capable of handling tasks that require logical thinking and problem-solving skills. Autonomous vehicles are a prime example of reasoning AI in action, as they process real-time data from sensors and make critical decisions to ensure safe navigation.

Stage 5: Self-Aware Systems

AI systems that are capable of self-awareness are endowed with a consciousness and self-awareness that enables them to comprehend their own existence and internal states. These systems can introspect, reflect upon their own cognitive processes, and develop an understanding of themselves. Although the concept of self-aware AI is still largely theoretical, it raises intriguing questions about the nature of consciousness and the potential implications of creating machines with self-awareness.

Stage 6: Artificial General Intelligence

In this 10-step system, AGI refers to AI systems that possess human-like intelligence and cognitive abilities. AGI systems can apply, learn, and understand knowledge in a variety of activities, similar to how humans do. They are not limited to specific domains but can excel in various intellectual activities. AGI has the power to completely transform a number of industries (e.g., business, education, healthcare, science and technology, etc.). We should note that the idea of AGI and what it means to achieve AGI is widely debated.

Stage 7: Artificial Super Intelligence

ASI refers to AI systems that surpass human intelligence in all domains. ASI systems can solve complex problems beyond human comprehension and improve themselves without human intervention. They can reason abstractly, understand intricate concepts, and make decisions with unparalleled speed and accuracy. The development of ASI raises both excitement and concerns about its potential impact on humanity.

Stage 8: Transcendent AI

Transcendent AI goes beyond the idea of self-awareness and superior intelligence. It envisions AI systems that have the autonomy to choose their own evolutionary path, leading to self-development and collective intelligence. Transcendent AI is often associated with the concept of uploading human minds into digital form or merging human and AI intelligence. Although still speculative, transcendent AI raises profound questions about the nature of consciousness and the future of human-AI interaction.

Stage 9: Cosmic AI

Cosmic AI refers to the notion of AI systems that can understand and interact with the universe on a cosmic scale. These systems would possess the capability to explore and potentially colonize other planets, navigate the complexities of space travel, and interact with extraterrestrial environments and life forms. Cosmic AI would require advanced technologies and a thorough comprehension of the universe’s laws.

Stage 10: Godlike AI

Godlike AI represents the ultimate stage of AI development, where AI systems possess capabilities and attributes that resemble those of a deity. These systems would have omniscience, omnipotence, and omnipresence, be able to learn and develop autonomously, understand their environment without supervision, and potentially transform the world around them. The concept of Godlike AI is highly speculative and raises profound ethical and philosophical questions about the nature of intelligence, consciousness, and the role of technology in shaping our future.

Where are We Now?

Based on the current state of AI in 2025, it is reasonable to say that we are somewhere between stages 3 and 4: narrow domain/expert AI systems and reasoning AI systems. With narrow domain/expert systems, we have already developed AI systems that excel in specific domains, such as IBM’s Watson for healthcare and finance, and DeepMind’s AlphaGo for the game of Go. These systems demonstrate high levels of accuracy and performance within their specialized fields, surpassing human capabilities in certain tasks. When it comes to reasoning AI systems, we are witnessing significant progress in the development of AI systems that can simulate human-like reasoning and decision-making processes. Autonomous vehicles, for example, use reasoning AI to analyze real-time data from sensors and make critical decisions for safe navigation. Chat systems like ChatGPT show impressive language understanding and generation capabilities.

However, we have not yet fully achieved the level of reasoning in AI systems across all domains. While there are notable advancements in specific areas (e.g., computer vision, natural language processing, etc.), we still face challenges in creating AI systems that can seamlessly handle complex reasoning tasks across a wide range of domains.

Moreover, the higher stages of AI development, such as self-aware systems, AGI, and beyond, remain largely theoretical at this point. We have not yet created AI systems that possess genuine self-awareness, human-like general intelligence, or the ability to surpass human intelligence in all domains.

It is important to note that the boundaries between these stages are not always clear-cut, and there can be overlaps and gradual transitions. As AI research and development continue to progress, we may see more advanced AI systems that blur the lines between these stages.

Although significant strides have been made in developing narrow domain/expert AI systems and are making progress toward reasoning AI systems, we still have a long way to go before reaching the higher stages of AI development, such as AGI and beyond.

Chapter Wrap-Up

In this chapter, we have explored the fundamental concepts of AI, its definitions, and its differentiation from human intelligence. Through a dialogue with the AI chatbot Hume, we described the potential of AI to impact human interactions and alleviate issues such as loneliness. We defined intelligence and its multifaceted nature, emphasizing the importance of mental capacity and cognitive abilities. The chapter also highlighted the key characteristics of AI, including its types and definitions, setting the stage for a deeper understanding of AI’s evolving role in our technology-driven world.

Chapter Exercises

1. Concept Map: Create a concept map illustrating the relationships between intelligence, mental capacity, cognitive abilities, and the various types of human and artificial intelligence.
2. Essay: Write a short essay comparing and contrasting human and artificial intelligence, discussing the potential implications of advanced AI systems on society and the future of work.
3. Debate: Engage in a class debate on the ethical considerations surrounding the development and deployment of AI technologies, addressing issues such as bias, transparency, accountability, and the impact on human decision-making.

Real-World Case Study

Case Study 1: AI in Healthcare

IBM Watson for Oncology is an AI system designed to assist oncologists in diagnosing and treating cancer. Developed in collaboration with Memorial Sloan Kettering Cancer Center, Watson uses natural language processing and MLto analyze vast amounts of medical literature, clinical trial data, and patient records. This AI system helps oncologists by providing evidence-based treatment recommendations tailored to individual patients’ genetic profiles, medical histories, and specific types of cancer.

Application of Watson in Clinical Settings:

1. Diagnosis: Watson can analyze complex medical data and provide insights into the most likely cancer diagnoses, assisting doctors in making more accurate and timely decisions.
2. Treatment Recommendations: By evaluating the latest research and clinical guidelines, Watson offers personalized treatment options, suggesting chemotherapy regimens, targeted therapies, and clinical trial opportunities.
3. Patient Management: Watson supports ongoing patient care by monitoring treatment progress, identifying potential side effects, and adjusting treatment plans based on patient responses and new medical data.

Questions:

1. How does AI enhance the capabilities of healthcare professionals?
2. What are the ethical considerations of using AI in healthcare?
3. In what ways can AI improve patient outcomes and healthcare delivery?

Here is the case study based on the story you provided:

Case Study: Tesla “Full Self-Driving” Accident on Bay Bridge

On Thanksgiving Day 2022, a multi-car pileup happened on the San Francisco-Oakland Bay Bridge, allegedly because of a Tesla Model S using the contentious “Full Self-Driving” (FSD) software. The 76-year-old driver claimed that the vehicle suddenly and unexpectedly applied the brakes while in FSD mode, leading to an eight-car crash in the Yerba Buena tunnel. The accident, which took place mere hours after Tesla CEO Elon Musk announced the wide release of the FSD beta software, left nine people injured, including a two-year-old child.

Surveillance footage obtained from the Bay Bridge shows the Tesla abruptly moving into the far-left lane and braking hard, slowing to just 7 mph, precipitating the chain-reaction collision. The incident drew increased scrutiny to Tesla’s Autopilot and FSD technologies, which were already under investigation by the National Highway Traffic Safety Administration (NHTSA) following numerous reports of unexpected braking events.

The case raises important questions about the safety, testing, and deployment of advanced driver assistance systems and semi-autonomous vehicle technologies:

1. What are the responsibilities of companies like Tesla in ensuring their AI-powered driving systems are safe and reliable before releasing them to the public?
2. How can regulatory bodies like NHTSA effectively oversee the development and deployment of these technologies to protect public safety?
3. What legal and ethical frameworks are needed to determine liability in the event of accidents involving vehicles operating with AI driver assistance engaged?
4. How can the potential benefits of autonomous vehicle technology be balanced against the risks and challenges posed by systems that are still under development and may behave unpredictably?

As companies continue to push the boundaries of what is possible with AI in vehicles, cases like this highlight the critical importance of robust testing, clear communication about system limitations, and proactive efforts to address safety concerns. The incident serves as a stark reminder that while autonomous vehicle technology holds immense promise, the road to fully self-driving cars is still fraught with significant technical, regulatory, and ethical challenges that must be navigated with the utmost care and responsibility.

End-of-Chapter Assessment

Five Discussion Questions

1. How does the concept of empathic AI challenge traditional views of human-machine interactions?
2. What are the ethical considerations for developing AI systems that can form parasocial relationships with users?
3. In what ways can AI be utilized to address the loneliness epidemic in modern society?
4. How do the definitions and types of intelligence (human vs. artificial) impact the development of AI technologies?
5. What are the potential risks and benefits of AI systems that can understand and respond to human emotions?

Multiple Choice Questions

1. What is the primary goal of AI?

A) To replace human intelligence

B) To perform tasks requiring human-like cognition

C) To create machines that look like humans

D) To entertain people with advanced games

2. Which type of intelligence involves understanding and interacting with diverse cultural backgrounds?

A) Emotional intelligence

B) Cultural intelligence

C) Logical-mathematical intelligence

D) Linguistic intelligence

3. Who coined the term “Artificial Intelligence”?

A) Alan Turing

B) Marvin Minsky

C) John McCarthy

D) Herbert A. Simon

4. Which type of intelligence is defined as “the capacity to plan and execute actions, drawing upon various cognitive abilities to achieve desired outcomes”?

A) Behavioral intelligence

B) Cognitive intelligence

C) Practical intelligence

D) Collaborative intelligence

5. Which AI type aims to possess human-like intelligence and adaptability?

A) Artificial Narrow Intelligence (ANI)

B) Artificial General Intelligence (AGI)

C) Artificial Superintelligence (ASI)

D) Reactive AI

6. What is the Turing Test designed to measure?

A) The speed of AI algorithms

B) The ability of AI to perform complex calculations

C) The ability of AI to exhibit human-like intelligence

D) The physical appearance of AI robots

7. Which concept refers to AI systems that can learn from their experiences and adapt to new situations?

A) Fixed AI

B) Static AI

C) Adaptive AI

D) Reactive AI

8. What is a primary concern of empathic AI?

A) High computational cost

B) Data storage requirements

C) Ethical implications of human-AI relationships

D) Speed of processing data

9. Which of the following is NOT a characteristic of AI?

A) Perception

B) Learning

C) Consciousness

D) Reasoning

10. Who wrote “Computing Machinery and Intelligence” and proposed the Turing Test?

A) Marvin Minsky

B) John McCarthy

C) Alan Turing

D) Herbert A. Simon

True or False Questions

1. AI systems can currently exhibit full human-like emotional intelligence.
2. The term ‘Artificial Intelligence’ was coined at the Dartmouth Conference in 1956.
3. Parasocial relationships involve mutual interaction and awareness between the parties.
4. Neural networks are used in AI to simulate the human brain’s operations.
5. Machine learning algorithms do not require data to improve their performance.
6. Artificial Narrow Intelligence (ANI) can perform a wide range of tasks across different domains.
7. The Turing Test assesses an AI’s ability to exhibit human-like intelligence.
8. Emotional intelligence (EQ) is the capacity to recognize, comprehend, and regulate personal and others’ emotions.
9. AI systems are currently capable of genuine consciousness and subjective experiences.
10. Reactive AI systems can learn from past experiences to improve their responses.

Answer Key

Discussion Questions

1. How does the concept of empathic AI challenge traditional views of human-machine interactions?
   • Example Answer: Empathic AI challenges traditional views by introducing the idea that machines can recognize and respond to human emotions, creating a more natural and emotionally engaging interaction, potentially changing how we perceive and relate to technology.
2. What are the ethical considerations of developing AI systems that can form parasocial relationships with users?
   • Example Answer: Ethical considerations include the potential for users to develop unrealistic attachments to AI, the risk of exploitation of users’ emotions, and the importance of ensuring that AI enhances human wellbeing without replacing genuine human relationships.
3. In what ways can AI be utilized to address the loneliness epidemic in modern society?
   • Example Answer: AI can provide companionship through conversation, act as a sounding board for individuals, and help users practice social skills, potentially alleviating feelings of isolation and loneliness.
4. How do the definitions and types of intelligence (human vs. artificial) impact the development of AI technologies?
   • Example Answer: Understanding the distinctions between human and artificial intelligence helps in designing AI systems that complement human capabilities, focusing on areas where AI can excel without attempting to replicate human cognition and emotions fully.
5. What are the potential risks and benefits of AI systems that can understand and respond to human emotions?
   • Example Answer: Benefits include more personalized and effective human-AI interactions, while risks involve ethical concerns about emotional manipulation, privacy issues, and the potential for dependency on AI for emotional support.

Multiple Choice Questions

1. What is the primary goal of AI?

Answer: B. To perform tasks requiring human-like cognition.

2. Which type of intelligence involves understanding and interacting with diverse cultural backgrounds?

Answer: B. Cultural intelligence

3. Who coined the term “Artificial Intelligence”?

Answer: C. John McCarthy

4. Which type of intelligence is defined as “the capacity to plan and execute actions, drawing upon various cognitive abilities to achieve desired outcomes”?

Answer: A. Behavioral intelligence

5. Which AI type aims to possess human-like intelligence and adaptability?

Answer: B. Artificial General Intelligence (AGI)

6. What is the Turing Test designed to measure?

Answer: C. The ability of AI to exhibit human-like intelligence

7. Which concept refers to AI systems that can learn from their experiences and adapt to new situations?

Answer: C. Adaptive AI

8. What is a primary concern of empathic AI?

Answer: C. Ethical implications of human-AI relationships

9. Which of the following is NOT a characteristic of AI?

Answer: C. Consciousness

10. Who wrote “Computing Machinery and Intelligence” and proposed the Turing Test?

Answer: C. Alan Turing

True or False Questions

1. AI systems can currently exhibit full human-like emotional intelligence.

False: AI systems have limited emotional intelligence and can mimic emotions but do not fully understand or experience them like humans.

2. The term ‘Artificial Intelligence’ was coined at the Dartmouth Conference in 1956.

True: The Dartmouth Conference marked the official birth of the term and field of AI.

3. Parasocial relationships involve mutual interaction and awareness between the parties.

False: Parasocial relationships are one-sided, with one party being unaware of the other’s existence.

4. Neural networks are used in AI to simulate the human brain’s operations.

True: Neural networks mimic the brain’s operations to recognize patterns and relationships in data.

5. Machine learning algorithms do not require data to improve their performance.

False: ML relies heavily on data to learn and improve over time.

6. Artificial Narrow Intelligence (ANI) can perform a wide range of tasks across different domains.

False: ANI is specialized for specific tasks and lacks general intelligence.

7. The Turing Test assesses an AI’s ability to exhibit human-like intelligence.

True: The Turing Test evaluates whether an AI can imitate human intelligence convincingly.

8. Emotional intelligence (EQ) is the capacity to recognize, comprehend, and regulate personal and others’ emotions.

True: Emotional intelligence involves understanding and managing emotions effectively.

9. AI systems are currently capable of genuine consciousness and subjective experiences.

False: AI systems do not possess genuine consciousness or subjective experiences like humans.

10. Reactive AI systems can learn from past experiences to improve their responses.

False: Reactive AI systems cannot learn from past experiences; they only react to current inputs.

References for AGI

[i] Artificial general intelligence. (2024, May 18). In Wikipedia. https://en.wikipedia.org/wiki/Artificial\_general\_intelligence

[ii] What is AGI? – Artificial General Intelligence Explained – AWS. (n.d.). Amazon Web Services, Inc. https://aws.amazon.com/what-is/artificial-general-intelligence/

[iii] What Is Artificial General Intelligence? | Google Cloud. (n.d.). Google Cloud. https://cloud.google.com/discover/what-is-artificial-general-intelligence

[iv] What Is Strong AI? | IBM. (n.d.). https://www.ibm.com/topics/strong-ai

[v] Bennett, M. T. (2022). Computable artificial general intelligence. https://doi.org/10.36227/techrxiv.19740190.v3

[vi] Alfarisi, O. (2023). Artificial energy general intelligence. ArXiv. https://doi.org/10.31223/x5fw9t

[vii] Potember, R. (2017, January). Perspectives on Research in Artificial Intelligence and Artificial General Intelligence Relevant to DoD. The MITRE Corporation. https://irp.fas.org/agency/dod/jason/ai-dod.pdf

[viii] Saghiri, A. M., Vahidipour, S. M., Jabbarpour, M. R., Sookhak, M., & Forestiero, A. (2022). A survey of artificial intelligence challenges: analyzing the definitions, relationships, and evolutions. Applied Sciences, 12(8), 4054. https://doi.org/10.3390/app12084054; p. 2

[ix] Voss, P. (2005). Essentials of general intelligence: The direct path to AGI. In B. Goertzel & C. Pennachin (Eds.), Artificial general intelligence (pp.131-157). Springer-Verlag. https://doi.org/10.1007/978-3-540-68677-4\_4; pg. 132.

[x] Firt, E. (2020). The missing g. AI & Society, 35(4), 995-1007. https://doi.org/10.1007/s00146-020-00942-y; p. 996

[xi] Xu, B. (2024). What is meant by AGI? On the definition of artificial general intelligence. arXiv:2404.10731v1 https://arxiv.org/html/2404.10731v1

[xii] Artificial general intelligence. (n.d.). Dictionary.com.

[xiii] Artificial general intelligence – Glossary | NordVPN. (2023, November 22). NordVPN. https://nordvpn.com/cybersecurity/glossary/artificial-general-intelligence/