Chapter 9
Artificial Intelligence
Fundamentals
Foundations
ANI, AGI, ASI
OVERVIEW
Artificial Narrow Intelligence (ANI), also called weak AI, refers to systems designed to perform a specific task or narrow set of tasks, operating within predefined parameters and unable to transfer learning beyond their domain. Every AI in existence today is ANI, including image recognition tools, language models, recommendation engines, and game-playing programs like Deep Blue and AlphaGo. Despite seeming sophisticated, these systems lack genuine understanding or consciousness, executing pattern recognition rather than true reasoning. Artificial General Intelligence (AGI), or strong AI, describes a hypothetical system possessing human-level cognitive abilities across virtually any intellectual task, capable of learning, reasoning, planning, and adapting knowledge flexibly between unrelated domains the way humans do. AGI would understand context, exhibit common sense, and solve novel problems without task-specific training. It does not yet exist. Artificial Super Intelligence (ASI) represents an even more speculative stage in which machine intelligence vastly surpasses the brightest human minds in essentially every field, including scientific creativity, social skills, and general wisdom. Theorists envision ASI as potentially self-improving, recursively enhancing its own capabilities at an accelerating pace. While ANI is present reality, AGI remains an aspirational research goal, and ASI exists purely as theoretical projection beyond AGI's eventual arrival.
INFLUENCES
The development of ANI traces to foundational figures including Alan Turing, whose 1950 work framed machine intelligence, and the 1956 Dartmouth Conference convened by John McCarthy, Marvin Minsky, and others that named the field. Progress accelerated through machine learning advances, the backpropagation algorithm refined by Geoffrey Hinton and colleagues, and the deep learning revolution catalyzed by AlexNet in 2012, alongside massive datasets and GPU computing power. AGI's conceptual influences include early cyberneticists, science fiction visionaries, and researchers like Ben Goertzel, who popularized the term, plus organizations such as DeepMind and OpenAI explicitly pursuing it. Cognitive science, neuroscience-inspired architectures, and the transformer model introduced by Google researchers in 2017 have shaped AGI ambitions. ASI's intellectual lineage stems largely from theorists rather than engineers. Mathematician I.J. Good articulated the "intelligence explosion" concept in 1965, describing recursive self-improvement. Philosopher Nick Bostrom's 2014 book Superintelligence profoundly influenced discourse around risks and control problems, while figures like Ray Kurzweil promoted singularity theories. Vernor Vinge popularized the technological singularity notion. These thinkers, alongside the AI safety community including researchers at the Future of Humanity Institute and Machine Intelligence Research Institute, frame ASI primarily as an existential consideration.
IMPACT
ANI delivers pervasive real-world applications: virtual assistants, fraud detection, medical imaging diagnostics, autonomous vehicle subsystems, content recommendation, and generative tools for text and images, profoundly impacting commerce, healthcare, and daily communication while raising concerns about bias, job displacement, and privacy. Its innovations directly fuel progress toward more advanced systems, with large language models demonstrating increasingly general-seeming capabilities. AGI, being hypothetical, has no current applications, but its anticipated impact would be transformative, potentially revolutionizing scientific research, automating intellectual labor across every sector, accelerating drug discovery, and reshaping economies and education. Anticipating AGI has already spurred subsequent innovations in alignment research, interpretability, and governance frameworks designed to ensure beneficial outcomes. ASI similarly has no applications, existing only as projection, yet its theorized impact dominates long-term discussions: proponents envision solutions to humanity's greatest challenges like disease, climate change, and poverty, while critics warn of catastrophic or existential risks if such intelligence proves misaligned with human values. ASI speculation has driven innovation in AI safety methodologies, value-alignment research, and international policy proposals. Collectively, these three tiers represent a continuum from today's practical tools toward speculative futures shaping research priorities, ethics, and investment.
Innovators
John McCarthy
BRIEF BIO
John McCarthy (1927–2011) was an American computer scientist and cognitive scientist widely regarded as one of the founding fathers of artificial intelligence. Born in Boston, Massachusetts, to immigrant parents active in labor and political movements, McCarthy displayed exceptional mathematical talent from an early age, teaching himself college-level mathematics before enrolling. He earned his bachelor's degree in mathematics from the California Institute of Technology in 1948, then completed his doctorate in mathematics at Princeton University in 1951 under the supervision of Solomon Lefschetz. After holding positions at Princeton, Stanford, Dartmouth, and MIT, McCarthy eventually settled at Stanford University, where he founded the Stanford Artificial Intelligence Laboratory in 1962 and remained for the rest of his career until retiring as professor emeritus. He famously coined the term "artificial intelligence" in 1955 when proposing the seminal Dartmouth Summer Research Project, the 1956 conference now considered the founding event of the AI field. Known for his deep commitment to formalizing human reasoning through logic and mathematics, McCarthy possessed a famously independent, sometimes contrarian intellect and a visionary outlook. He devoted his life to the conviction that human intelligence could be precisely described and ultimately replicated by machines, pursuing this ambitious goal across more than five decades of pioneering research.
ACCOMPLISHMENTS
McCarthy's accomplishments span foundational theory and practical innovation across computing. He coined the term "artificial intelligence" and organized the 1956 Dartmouth Conference that launched the field as a formal discipline. In 1958 he invented Lisp, which became the dominant programming language for AI research and introduced influential concepts including recursion, conditional expressions, and automatic garbage collection that shaped modern programming. He pioneered the concept of time-sharing computer systems, allowing multiple users to access a single machine simultaneously, work that fundamentally influenced the development of interactive computing and later networked systems and utility computing. He proposed the idea of computation provided as a public utility, anticipating cloud computing decades before its realization. McCarthy developed circumscription, a formal method of non-monotonic reasoning addressing how machines handle incomplete information and commonsense reasoning. He founded artificial intelligence laboratories at both MIT, co-founding it with Marvin Minsky, and Stanford. His 1969 paper with Patrick Hayes introduced the situation calculus and articulated the influential frame problem. For these contributions he received the prestigious Turing Award in 1971, the National Medal of Science in 1990, the Kyoto Prize in 1988, and the Benjamin Franklin Medal, cementing his stature among computing's most decorated pioneers.
IMPACT
McCarthy's impact on computer science and society remains profound and enduring. By naming and conceptually framing artificial intelligence, he established an entire field of inquiry that now influences nearly every domain of modern life, from medicine and transportation to communication and scientific discovery. Lisp, his programming language, shaped generations of computer scientists and introduced programming paradigms still embedded in contemporary languages, while inspiring functional programming approaches that experience renewed relevance today. His advocacy for time-sharing fundamentally transformed how humans interact with computers, helping move computing away from isolated batch processing toward the interactive, accessible model underpinning personal and cloud computing. His prescient vision of computation as a utility anticipated the cloud-based infrastructure now central to the global digital economy. McCarthy's insistence on formal logic and mathematical rigor established methodological standards influencing AI research for decades and laid groundwork for knowledge representation, automated reasoning, and the symbolic AI tradition. Beyond technical contributions, he trained and mentored numerous students who became leaders in computing, propagating his influence across the discipline. Though he did not live to witness recent breakthroughs in machine learning, the field he founded continues expanding dramatically, ensuring his intellectual legacy persists as artificial intelligence increasingly reshapes humanity's future and possibilities.
Events
Deep Blue
OVERVIEW
Deep Blue was a chess-playing supercomputer developed by IBM that became the first computer system to defeat a reigning world chess champion under standard tournament conditions, achieving this historic milestone against Garry Kasparov in 1997. It exemplified the reactive machine paradigm of artificial intelligence, evaluating the current board position and calculating optimal moves without learning from past games or retaining memory between matches. Deep Blue relied primarily on brute-force computational power combined with sophisticated chess knowledge, employing a massively parallel architecture capable of examining approximately two hundred million chess positions per second. The system used a specialized search algorithm based on the minimax method with alpha-beta pruning, exploring possible move sequences many plies deep into the game tree to identify advantageous positions. Custom-designed chess chips, with hundreds working in parallel, accelerated this enormous computation, while an evaluation function encoded chess principles assessing factors like material balance, king safety, piece mobility, and pawn structure. Expert chess knowledge from grandmasters, including consultant Joel Benjamin, was incorporated to refine its positional judgment and opening strategies. Rather than exhibiting genuine understanding or human-like reasoning, Deep Blue demonstrated that raw computational speed paired with well-tuned heuristics could conquer a domain long considered a quintessential test of human intelligence and strategic thinking.
INFLUENCES
Deep Blue's development drew from decades of computer chess research and pioneering theoretical work. Claude Shannon's seminal 1950 paper outlining how a computer might play chess established foundational principles, including the minimax search strategy and position evaluation functions that guided subsequent efforts. Alan Turing's early chess algorithm conceived around the same period further seeded the field. The project originated from a Carnegie Mellon University effort called ChipTest, developed by graduate student Feng-hsiung Hsu, which evolved into Deep Thought, a precursor system that Hsu and colleagues including Murray Campbell and Thomas Anandan built. When IBM recruited this team in 1989, the project was renamed and substantially expanded with corporate resources. Feng-hsiung Hsu designed the specialized chess chips central to Deep Blue's processing power, while Murray Campbell and Joseph Hoane contributed to the software and system architecture. Grandmaster Joel Benjamin assisted in tuning the evaluation function with sophisticated chess expertise. The earlier 1996 match, in which Kasparov defeated an initial version of Deep Blue, provided crucial lessons that informed substantial upgrades to the system's speed and chess knowledge before the triumphant 1997 rematch. Advances in parallel computing, custom chip fabrication, and search algorithm optimization collectively enabled the machine's unprecedented analytical capabilities.
IMPACT
Deep Blue's primary application was competitive chess, yet its significance extended far beyond the game itself, marking a watershed moment in artificial intelligence history. Its 1997 victory demonstrated that machines could surpass human experts in domains demanding deep strategic calculation, profoundly impacting public perception of computing's potential and igniting widespread discussion about machine intelligence and its implications for human cognition. The achievement validated brute-force search combined with domain expertise as a powerful problem-solving approach, influencing subsequent work in optimization, decision-making, and search-intensive computational problems across fields like logistics, finance, and scheduling. The specialized parallel processing techniques and high-performance computing architectures developed for Deep Blue informed IBM's broader research, contributing to later supercomputing initiatives. IBM subsequently pursued more ambitious challenges, developing Watson, which triumphed at the quiz show Jeopardy! in 2011, demonstrating natural language understanding, and later advancing systems exploring practical applications in healthcare and business analytics. Deep Blue also helped clarify a crucial distinction in AI, illustrating that defeating humans at chess required narrow, specialized capability rather than general intelligence or genuine understanding. This insight shaped expectations and spurred pursuit of more flexible, learning-based approaches, ultimately influencing the trajectory toward machine learning and the neural network methods dominating contemporary artificial intelligence research and applications today.
Companies
COMPANY PROFILE
Google has played a central and arguably unparalleled role in the development of modern artificial intelligence, functioning as both a foundational research powerhouse and a major deployer of AI technologies across consumer products used by billions. The company's involvement spans fundamental theoretical breakthroughs, open-source software tools, specialized hardware, and large-scale practical applications. Google's research division and its DeepMind subsidiary, acquired in 2014, have produced numerous landmark advances, most notably the transformer architecture introduced in the 2017 paper "Attention Is All You Need," which became the foundational design underpinning virtually all contemporary large language models, including those powering competing systems. Google developed and open-sourced TensorFlow, a widely adopted machine learning framework that democratized deep learning development worldwide, and engineered custom Tensor Processing Units, specialized chips accelerating AI computation. The company has applied AI extensively throughout its product ecosystem, enhancing search, translation, photos, advertising, and recommendation systems. Through DeepMind, Google achieved historic milestones including AlphaGo's defeat of world Go champions and AlphaFold's revolutionary protein structure predictions. Google also developed its own conversational AI systems and large language models, including the Gemini family, positioning the company simultaneously as a pioneering researcher advancing the scientific frontier and a dominant commercial force shaping how artificial intelligence reaches and affects everyday users globally.
INFLUENCES
Google's AI development drew upon and amplified numerous influences across the field's history. The company recruited many of the discipline's leading minds, including Geoffrey Hinton, whose neural network startup Google acquired in 2013, bringing foundational deep learning expertise in-house. Its acquisition of DeepMind, co-founded by Demis Hassabis, integrated a world-class research organization pursuing artificial general intelligence through reinforcement learning and neuroscience-inspired approaches. The broader deep learning revolution catalyzed by AlexNet in 2012 demonstrated neural networks' potential, encouraging Google's substantial investment in the technology. Google Brain, the company's pioneering deep learning research team founded by Andrew Ng, Jeff Dean, and others, advanced large-scale neural network training and produced influential work, including the transformer architecture developed by researchers including Ashish Vaswani and colleagues. The availability of Google's enormous datasets from search, video, and other services provided unparalleled training material, while its massive computational infrastructure supplied necessary processing power. Academic traditions in machine learning, statistics, and computer science informed its methods, and competitive pressure from rivals like OpenAI accelerated development. Jeff Dean's engineering leadership in building scalable systems proved instrumental. The convergence of exceptional talent, abundant data, computational resources, and institutional commitment positioned Google at the forefront of artificial intelligence research and innovation.
IMPACT
Google's AI applications permeate daily life globally, powering search ranking, Google Translate across hundreds of languages, voice recognition in Google Assistant, image organization in Google Photos, spam filtering in Gmail, and recommendation systems throughout YouTube and advertising platforms, affecting billions of users continuously. The company's impact on artificial intelligence as a field is profound, since its transformer architecture fundamentally enabled the entire generative AI revolution, including large language models that now define the technology's frontier. By open-sourcing TensorFlow and publishing extensive research, Google accelerated worldwide AI progress and education, though it also faced criticism regarding the concentration of power, data privacy, and the ethics of deploying AI at scale. DeepMind's scientific breakthroughs, particularly AlphaFold, demonstrated AI's transformative potential for accelerating fundamental research, earning Nobel recognition. Subsequent innovations from Google include the Gemini family of multimodal large language models, advances in AI-powered medical diagnostics, weather forecasting systems, and materials discovery. The company integrated generative AI throughout its products, embedding assistants into search and workplace tools. Google's responses to competitive pressure spurred rapid deployment of conversational AI, though not without challenges navigating accuracy, safety, and responsible development. As both researcher and commercial titan, Google continues shaping artificial intelligence's trajectory, balancing scientific ambition, product integration, and growing scrutiny over its expansive influence across the global technology landscape.