THE PIONEERS OF MACHINE SUPERINTELLIGENCE

From Symbolic Reasoning to Adaptive Neural Intelligence

The development of Machine Superintelligence has also been shaped by changing philosophical perspectives concerning intelligence itself. Earlier research frequently sought to automate logical reasoning through symbolic computation, whereas later developments increasingly emphasised statistical learning, adaptive optimisation and biologically inspired neural architectures. Contemporary thinking increasingly integrates these perspectives, recognising that advanced intelligence requires continual interaction among reasoning, learning, perception, memory, planning and creativity. Consequently, the pioneers of Machine Superintelligence should be viewed collectively rather than individually, each contributing essential intellectual components that continue influencing research into increasingly sophisticated forms of Artificial Intelligence.

This white paper examines the principal pioneers whose ideas collectively established the intellectual foundations of Machine Superintelligence. It explores their contributions, evaluates their continuing influence and demonstrates how successive generations of scientific innovation have progressively shaped contemporary understanding of advanced computational intelligence.

Machine Superintelligence as Integrated and Adaptive Cognition

Machine Superintelligence may be defined as a future form of Artificial Intelligence whose intellectual capabilities consistently exceed those of the most capable human experts across virtually every cognitive discipline. Such capability extends beyond specialised computational performance to include scientific reasoning, mathematical discovery, creative innovation, strategic planning, interdisciplinary synthesis, autonomous learning and continual self-improvement. Unlike existing Artificial Intelligence systems, which remain largely specialised within defined operational domains, Machine Superintelligence implies comprehensive intellectual capability characterised by adaptability, autonomy and integrated cognition.

The concept has evolved gradually through decades of scientific investigation into the nature of intelligence. Early pioneers sought to determine whether reasoning itself could be represented mathematically and executed computationally. Subsequent researchers expanded these investigations by examining learning, adaptation, knowledge representation and probabilistic inference. More recent developments have focused upon scalable neural computation, multimodal learning, reinforcement learning and increasingly autonomous cognitive architectures. Each stage represents an incremental extension of earlier theoretical foundations while simultaneously introducing new perspectives concerning computational intelligence.

Machine Superintelligence therefore represents not merely an engineering objective but the culmination of numerous intellectual traditions. Its foundations lie equally within mathematics, philosophy, neuroscience, engineering and computer science, making its pioneers collectively responsible for one of the most significant scientific enterprises of the modern era.

Logic, Computation, Cybernetics and Information Theory

The intellectual origins of Machine Superintelligence predate the formal establishment of Artificial Intelligence as an academic discipline. During the first half of the twentieth century, developments in mathematical logic, formal computation and information theory fundamentally altered scientific understanding of reasoning and information processing. These developments established the theoretical conditions necessary for later exploration of intelligent computation.

Mathematical logic demonstrated that reasoning could be represented through formal symbolic systems governed by precise rules. Simultaneously, advances in computation established that abstract logical procedures could be implemented mechanically through programmable machines. Together these discoveries transformed intelligence from an exclusively philosophical concept into an increasingly scientific subject capable of computational investigation.

The Second World War further accelerated these developments through rapid advances in cryptography, communications engineering, operations research and electronic computation. Scientists recognised that computational systems could perform tasks previously requiring considerable human intellectual effort, stimulating broader investigation into the possibility of automated reasoning and intelligent decision-making.

The emergence of cybernetics introduced another important intellectual influence by examining communication, control and feedback within biological and mechanical systems. Cybernetic theory demonstrated that adaptive behaviour could emerge through continual interaction between systems and their environments, foreshadowing later developments in machine learning and autonomous Artificial Intelligence. Information theory simultaneously established rigorous mathematical principles governing communication, uncertainty and knowledge representation, providing essential foundations for modern computational learning.

These intellectual developments collectively established the conceptual environment within which Artificial Intelligence emerged during the mid-twentieth century. Although Machine Superintelligence remained many decades away as a formal concept, its essential theoretical foundations had already begun taking shape through interdisciplinary scientific collaboration.

Foundational Pioneers of Computational Intelligence

Alan Turing: Universal Computation and Machine Intelligence

Alan Turing occupies a uniquely influential position within the intellectual history of Machine Superintelligence because he fundamentally transformed understanding of computation itself. His theoretical description of universal computation established that sufficiently general computational machines could execute any formally describable algorithm, providing the conceptual basis for programmable computers and, ultimately, Artificial Intelligence.

Turing's influence extends well beyond computational theory. His celebrated paper Computing Machinery and Intelligence introduced one of the earliest systematic examinations of whether machines might exhibit intelligent behaviour comparable with human reasoning. Rather than becoming preoccupied with abstract philosophical definitions of intelligence, Turing proposed evaluating intelligent behaviour operationally through conversational interaction. Although the Turing Test no longer represents a complete measure of intelligence, it initiated serious scientific discussion concerning machine cognition and established Artificial Intelligence as a legitimate field of academic enquiry.

Equally significant was Turing's anticipation that machines might acquire intelligence through learning rather than explicit programming. He proposed educational approaches through which computational systems could gradually develop increasingly sophisticated capabilities, foreshadowing many principles underlying contemporary machine learning. His vision of adaptive computational intelligence continues influencing discussions surrounding Machine Superintelligence, particularly regarding autonomous learning and continual intellectual development.

John von Neumann: Architecture, Game Theory and Self-Reproducing Systems

John von Neumann contributed profoundly to the conceptual foundations of Machine Superintelligence through his work on computer architecture, game theory, mathematical optimisation and self-reproducing automata. His architectural principles established the fundamental organisation of modern digital computers, creating computational platforms capable of supporting increasingly sophisticated Artificial Intelligence.

Perhaps even more influential was von Neumann's exploration of self-replicating computational systems. He recognised that sufficiently advanced computational machines might possess the capacity to reproduce and modify themselves according to formal mathematical principles. Although originally developed within theoretical biology and automata theory, these ideas later influenced discussions concerning recursive self-improvement, one of the defining characteristics frequently associated with Machine Superintelligence.

Von Neumann also recognised the accelerating pace of technological progress, suggesting that cumulative scientific development might fundamentally transform civilisation. His observations concerning technological acceleration continue informing contemporary debate regarding the long-term implications of increasingly capable Artificial Intelligence.

Norbert Wiener: Cybernetics, Feedback and Social Responsibility

Norbert Wiener established cybernetics, creating one of the earliest comprehensive scientific frameworks for understanding communication, control and adaptive behaviour within both biological organisms and computational systems. Cybernetics fundamentally altered conceptions of intelligence by demonstrating that purposeful behaviour could emerge through continual feedback rather than predetermined instruction alone.

Wiener's emphasis upon feedback, adaptation and self-regulation directly anticipated later developments in reinforcement learning, autonomous systems and adaptive optimisation. Machine Superintelligence depends heavily upon these principles because advanced intelligence requires continual interaction with changing environments rather than static computational behaviour.

Equally important was Wiener's recognition of the broader societal consequences associated with intelligent automation. Long before contemporary discussions concerning Artificial Intelligence governance, he warned that increasingly capable computational systems would influence employment, economics and social organisation. His ethical perspective established an enduring tradition linking technological progress with broader societal responsibility.

Claude Shannon: Information Theory and Strategic Computation

Claude Shannon transformed scientific understanding of information itself through the development of information theory. By establishing rigorous mathematical principles governing communication, uncertainty and information transmission, he created theoretical foundations that remain indispensable throughout contemporary Artificial Intelligence.

Information theory influences Machine Superintelligence in numerous ways. Efficient learning depends upon effective representation and compression of information, while probabilistic reasoning relies heavily upon concepts such as entropy and uncertainty. Shannon's work therefore underpins many computational methodologies employed within modern machine learning, neural computation and probabilistic inference.

Shannon also explored machine chess and computational problem-solving, demonstrating early interest in intelligent algorithms capable of strategic reasoning. His investigations illustrated that computational systems might eventually address problems requiring planning, evaluation and decision-making rather than simple numerical calculation. Consequently, Shannon should be recognised not only as the founder of information theory but also as an important intellectual precursor to advanced Artificial Intelligence.

Early Architects of Artificial Intelligence

Marvin Minsky: Cognitive Agents and the Society of Mind

Marvin Minsky occupies a central position among the modern architects of Machine Superintelligence because he consistently argued that intelligence should be understood as the coordinated interaction of numerous specialised cognitive processes rather than a single unified mechanism. As a co-founder of the Artificial Intelligence Laboratory at the Massachusetts Institute of Technology, he helped establish Artificial Intelligence as a rigorous scientific discipline while encouraging interdisciplinary collaboration among computer science, psychology, mathematics and neuroscience.

Minsky's theory of the Society of Mind proposed that intelligence emerges through cooperation among multiple relatively simple cognitive agents, each performing specialised functions while collectively producing sophisticated intellectual behaviour. This perspective remains highly relevant to Machine Superintelligence because contemporary research increasingly views advanced Artificial Intelligence as an integrated architecture combining reasoning, memory, perception, planning and learning rather than relying upon isolated computational algorithms.

His broader philosophical contributions challenged simplistic interpretations of intelligence and encouraged researchers to investigate creativity, common-sense reasoning, abstraction and self-reflection as essential components of advanced cognition. Many contemporary discussions concerning cognitive architectures, modular intelligence and integrated reasoning continue reflecting intellectual themes first articulated by Minsky.

John McCarthy: General Artificial Intelligence and Symbolic Reasoning

John McCarthy was instrumental in defining Artificial Intelligence as a distinct academic discipline. His formal introduction of the term Artificial Intelligence at the Dartmouth Summer Research Project in 1956 provided both intellectual identity and scientific direction for the emerging field. More importantly, McCarthy consistently advocated the pursuit of general computational intelligence rather than narrowly specialised applications, a vision closely aligned with contemporary thinking surrounding Machine Superintelligence.

McCarthy made substantial contributions to symbolic reasoning, knowledge representation and formal logical systems, believing that intelligent behaviour required explicit manipulation of abstract concepts rather than statistical association alone. While later developments introduced complementary learning paradigms, symbolic reasoning continues influencing research into explainable Artificial Intelligence, causal reasoning and neuro-symbolic architectures.

His vision of computational systems capable of autonomous reasoning, common-sense knowledge and continual intellectual development established enduring research objectives that remain central to the long-term aspiration of Machine Superintelligence.

Architects of Computational Problem Solving and Cognition

Herbert Simon: Problem Solving and Bounded Rationality

Herbert Simon profoundly influenced the intellectual development of Machine Superintelligence by demonstrating that reasoning, decision-making and problem-solving could be examined scientifically rather than philosophically. His interdisciplinary work united economics, psychology, organisational theory and computer science, establishing many of the conceptual foundations upon which modern Artificial Intelligence continues to develop.

Together with Allen Newell, Simon developed the Logic Theorist and the General Problem Solver, among the earliest computational systems designed to imitate aspects of human reasoning. Although primitive by contemporary standards, these systems demonstrated that complex intellectual activities could be represented computationally through formal symbolic manipulation. Their work established that intelligent behaviour need not remain uniquely biological but could emerge through carefully designed computational architectures.

Simon also introduced the influential concept of bounded rationality, recognising that intelligent agents operate within practical limitations concerning information, computational resources and time. This insight remains highly relevant to Machine Superintelligence because future Artificial Intelligence must continually balance optimal reasoning with computational efficiency while operating within dynamic and uncertain environments. Rather than assuming unlimited rationality, Simon emphasised adaptive decision-making, an idea that continues to influence contemporary research into autonomous reasoning and intelligent planning.

Beyond his technical achievements, Simon consistently argued that understanding intelligence required collaboration across multiple scientific disciplines. This interdisciplinary perspective remains central to Machine Superintelligence, whose future development depends upon integrating computer science with neuroscience, mathematics, psychology, engineering and philosophy.

Allen Newell: Unified Cognitive Architectures

Allen Newell occupies an equally significant position within the intellectual history of Machine Superintelligence through his pioneering work on cognitive architectures and computational models of reasoning. His collaboration with Herbert Simon established some of the earliest demonstrations that formal computational systems could perform tasks previously associated exclusively with human intelligence.

Newell's enduring contribution lies in his pursuit of unified theories of cognition. Rather than constructing isolated computational systems for specific problems, he sought comprehensive architectures capable of supporting multiple forms of intelligent behaviour within a single coherent framework. This aspiration anticipated contemporary efforts to develop increasingly general forms of Artificial Intelligence that integrate learning, memory, reasoning and planning within unified computational systems.

His development of the Soar cognitive architecture further illustrated this commitment to integrated intelligence. Soar attempted to model how diverse cognitive processes cooperate continuously to support complex reasoning and adaptive behaviour. Although subsequent Artificial Intelligence research expanded beyond symbolic architectures, Newell's emphasis upon cognitive integration continues influencing discussions surrounding Machine Superintelligence.

Perhaps most importantly, Newell argued that understanding intelligence required identifying the underlying architectural principles common to all intelligent systems rather than concentrating solely upon isolated computational techniques. This perspective continues shaping research into general computational cognition and increasingly comprehensive Artificial Intelligence.

The Pioneers of the Deep Learning Revolution

Geoffrey Hinton: Representation Learning and Deep Neural Networks

Geoffrey Hinton transformed the trajectory of Artificial Intelligence by demonstrating the extraordinary potential of artificial neural networks for learning complex representations from extensive information. During periods when symbolic reasoning dominated Artificial Intelligence research, Hinton continued investigating neural computation, believing that learning represented a more promising route towards advanced intelligence than manually encoding explicit knowledge.

His work on back-propagation, distributed representation and deep neural architectures established many of the computational principles underlying contemporary machine learning. These advances enabled Artificial Intelligence to achieve unprecedented capability within language processing, image recognition, speech understanding and scientific modelling, thereby reshaping the entire field.

Hinton's research also contributed fundamentally to understanding hierarchical representation learning, whereby increasingly abstract conceptual structures emerge automatically through successive computational layers. Such capability represents one of the defining characteristics expected within Machine Superintelligence because advanced reasoning depends upon progressively richer internal representations rather than isolated statistical associations.

Beyond individual technical contributions, Hinton demonstrated remarkable intellectual persistence by continuing neural computation research despite decades of scepticism. His eventual success transformed Artificial Intelligence and established deep learning as one of the principal technological foundations from which future Machine Superintelligence may ultimately emerge.

Yoshua Bengio: Abstract Representation and Responsible Artificial Intelligence

Yoshua Bengio has played a central role in advancing the theoretical understanding of deep learning while consistently emphasising the importance of developing increasingly general and trustworthy forms of Artificial Intelligence. His research has addressed representation learning, generative modelling, attention mechanisms and continual learning, contributing directly to many capabilities now associated with advanced computational intelligence.

Bengio has been particularly influential in demonstrating that Artificial Intelligence should construct increasingly abstract conceptual representations rather than relying exclusively upon manually engineered features. These hierarchical representations enable systems to transfer knowledge across domains, adapt to unfamiliar situations and acquire progressively richer understanding from large-scale information.

Equally significant has been Bengio's sustained emphasis upon the long-term societal implications of increasingly capable Artificial Intelligence. He has consistently advocated responsible governance, transparency and safety research alongside technical innovation, recognising that progress towards Machine Superintelligence requires ethical responsibility as well as computational sophistication.

His work illustrates how theoretical innovation and societal awareness can develop together, reinforcing the principle that future Machine Superintelligence must combine extraordinary capability with responsible design and deployment.

Yann LeCun: Convolutional Networks and Self-Supervised Learning

Yann LeCun occupies a pivotal position within the development of modern Artificial Intelligence through his pioneering work on convolutional neural networks and efficient representation learning. His research demonstrated that computational systems could automatically identify increasingly sophisticated patterns within complex visual information, establishing techniques that later expanded into numerous domains beyond computer vision.

LeCun's contributions extend beyond image recognition. His work has consistently promoted self-supervised learning as a mechanism through which Artificial Intelligence may acquire extensive knowledge without exhaustive human annotation. Such approaches closely align with future Machine Superintelligence because genuinely autonomous intelligence requires continual learning from naturally occurring information rather than dependence upon manually prepared datasets.

His broader vision also emphasises embodied intelligence, world modelling and predictive reasoning, encouraging Artificial Intelligence to develop increasingly comprehensive internal models of physical reality. These ideas suggest pathways through which future Machine Superintelligence may integrate perception, reasoning and planning within coherent cognitive architectures capable of operating effectively across diverse environments.

Collectively, Hinton, Bengio and LeCun established the deep learning revolution that fundamentally transformed Artificial Intelligence during the early twenty-first century, providing essential computational foundations for increasingly sophisticated forms of Machine Superintelligence.

Contemporary Thinkers in General Intelligence and Governance

Stuart Russell: Alignment and Human-Compatible Artificial Intelligence

Stuart Russell has become one of the foremost contemporary authorities examining the long-term implications of increasingly capable Artificial Intelligence. While contributing extensively to technical Artificial Intelligence research, Russell has placed particular emphasis upon alignment, arguing that future intelligent systems must remain consistently compatible with legitimate human objectives.

His work reframed discussions surrounding Machine Superintelligence by demonstrating that increasing computational capability alone cannot guarantee beneficial outcomes. Instead, he argues that advanced Artificial Intelligence must incorporate uncertainty regarding human preferences while remaining continually receptive to human guidance. These principles have significantly influenced contemporary research concerning Artificial Intelligence safety, verification and governance.

Russell's textbook, Artificial Intelligence: A Modern Approach, co-authored with Peter Norvig, has educated generations of researchers and remains one of the most influential works within the discipline. Through both technical scholarship and policy engagement, Russell has helped shape contemporary understanding of responsible progress towards Machine Superintelligence.

Nick Bostrom: Superintelligence, Risk and Long-Term Strategy

Nick Bostrom occupies a distinctive position among the pioneers of Machine Superintelligence because he systematically explored its philosophical, strategic and societal implications. His landmark work Superintelligence: Paths, Dangers, Strategies transformed discussion of advanced Artificial Intelligence from speculative philosophy into rigorous interdisciplinary analysis.

Bostrom examined how recursive self-improvement, technological acceleration and increasing computational capability might eventually produce forms of intelligence substantially exceeding human cognition. Although some aspects remain theoretical, his framework has profoundly influenced contemporary debate regarding Artificial Intelligence governance, existential risk and long-term strategic planning.

Perhaps most importantly, Bostrom demonstrated that preparation for Machine Superintelligence should begin before its technological realisation. His emphasis upon proactive governance, alignment research and international cooperation continues shaping academic, governmental and industrial approaches to advanced Artificial Intelligence.

Shane Legg: Universal Intelligence and General Learning Systems

Shane Legg has contributed significantly to the theoretical understanding of general intelligence through his work on formal definitions of intelligence and machine learning. His research sought rigorous mathematical descriptions capable of evaluating intelligence independently of biological or technological implementation.

Legg's collaboration with Marcus Hutter produced influential theoretical models describing universal intelligence through computational and information-theoretic principles. These investigations established important conceptual foundations for evaluating increasingly general forms of Artificial Intelligence while clarifying distinctions between specialised capability and comprehensive intelligence.

As a co-founder of DeepMind, Legg has also contributed directly to practical Artificial Intelligence research, helping shape one of the world's leading organisations investigating increasingly capable learning systems. His work illustrates the productive relationship between theoretical investigation and technological implementation within the continuing evolution of Machine Superintelligence.

Demis Hassabis: Neuroscience, General Intelligence and Scientific Discovery

Demis Hassabis has played a transformative role in contemporary Artificial Intelligence by combining insights from neuroscience, cognitive psychology and computer science to pursue increasingly general forms of computational intelligence. As co-founder and Chief Executive of DeepMind, he has overseen research producing landmark achievements including AlphaGo, AlphaFold and numerous advances in reinforcement learning and scientific discovery.

Hassabis consistently advocates an interdisciplinary approach to Machine Superintelligence, arguing that understanding biological intelligence provides valuable guidance for developing advanced Artificial Intelligence. His research emphasises memory systems, planning, imagination and world modelling as essential components of increasingly general computational cognition.

Under his leadership, Artificial Intelligence has demonstrated unprecedented capability within scientific research, particularly through advances in protein structure prediction that have accelerated biological investigation worldwide. These achievements illustrate how Artificial Intelligence may evolve from computational assistance towards active scientific collaboration, a defining characteristic frequently associated with Machine Superintelligence.

Ilya Sutskever: Large-Scale Neural Models and Artificial Intelligence Safety

Ilya Sutskever has been one of the principal architects of contemporary large-scale neural language models and generative Artificial Intelligence. His work on recurrent neural networks, sequence modelling and transformer-based architectures has substantially influenced the rapid evolution of language understanding and generative reasoning.

Sutskever has consistently argued that sufficiently advanced learning systems may acquire increasingly general intellectual capability through exposure to extensive information and continual optimisation. His research has therefore contributed directly to contemporary understanding of scaling laws, representation learning and emergent computational behaviour, concepts that occupy central positions within discussions surrounding Machine Superintelligence.

Equally important has been his recognition that increasing capability must be accompanied by corresponding advances in safety, alignment and responsible governance. His influence extends beyond technical innovation towards broader consideration of how Machine Superintelligence should develop within frameworks promoting long-term societal benefit.

The Collective and Interdisciplinary Legacy

Although each pioneer contributed distinctive insights, their greatest significance lies in the cumulative intellectual tradition they collectively established. Turing demonstrated that computation could support intelligent behaviour. Von Neumann provided architectural and mathematical foundations. Wiener and Shannon established principles governing information, communication and adaptive systems. Minsky, McCarthy, Simon and Newell transformed Artificial Intelligence into a rigorous scientific discipline while exploring the architecture of cognition. Hinton, Bengio and LeCun revolutionised computational learning through neural computation. Russell, Bostrom, Legg, Hassabis and Sutskever have subsequently extended these foundations towards increasingly general, autonomous and responsibly governed forms of Artificial Intelligence.

Cumulative Innovation Across Scientific Generations

This progression illustrates that Machine Superintelligence has not emerged through isolated discoveries but through continuous interdisciplinary collaboration extending across mathematics, engineering, psychology, neuroscience, philosophy and computer science. Each generation inherited conceptual foundations from its predecessors while expanding scientific understanding in response to emerging computational possibilities. Consequently, the history of Machine Superintelligence is fundamentally a history of cumulative intellectual integration.

A Century of Foundations for Machine Superintelligence

Machine Superintelligence represents one of the most ambitious scientific objectives ever conceived, yet its conceptual foundations have been constructed progressively through the work of numerous pioneers spanning almost a century of intellectual development. From Alan Turing's universal computation and John von Neumann's architectural innovations to Norbert Wiener's cybernetics and Claude Shannon's information theory, the earliest pioneers established principles that continue underpinning modern Artificial Intelligence. Their work transformed abstract philosophical questions concerning intelligence into rigorous scientific investigation.

Subsequent generations expanded these foundations by exploring symbolic reasoning, cognitive architectures and computational problem-solving. Marvin Minsky, John McCarthy, Herbert Simon and Allen Newell established Artificial Intelligence as an interdisciplinary scientific discipline while investigating the structure of reasoning itself. Their intellectual ambition anticipated many characteristics now associated with Machine Superintelligence, particularly the integration of diverse cognitive capabilities within unified computational systems.

The deep learning revolution initiated by Geoffrey Hinton, Yoshua Bengio and Yann LeCun fundamentally altered the technological trajectory of Artificial Intelligence by demonstrating the extraordinary potential of representation learning, neural computation and large-scale optimisation. Contemporary thinkers including Stuart Russell, Nick Bostrom, Shane Legg, Demis Hassabis and Ilya Sutskever have subsequently expanded the field towards increasingly general intelligence while emphasising safety, governance and responsible development.

Collectively, these pioneers illustrate that Machine Superintelligence should be understood not as a single technological breakthrough but as the cumulative product of successive scientific generations united by a common aspiration to understand and reproduce intelligence itself. Their contributions continue shaping research into learning, reasoning, creativity, autonomy and cognitive integration while guiding contemporary exploration of increasingly capable forms of Artificial Intelligence. As future advances continue building upon these intellectual foundations, the legacy of these pioneers will remain central to understanding both the scientific evolution and the societal significance of Machine Superintelligence.

Bibliography

• Bostrom, N. Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press, 2014.
• Hinton, G., Bengio, Y. and LeCun, Y. ‘Deep Learning’, Nature, 521 (2015), pp. 436-444.
• McCarthy, J. ‘Programs with Common Sense’, in Mechanisation of Thought Processes. London: Her Majesty's Stationery Office, 1959.
• Minsky, M. The Society of Mind. New York: Simon and Schuster, 1986.
• Newell, A. Unified Theories of Cognition. Cambridge, MA: Harvard University Press, 1990.
• Russell, S. Human Compatible: Artificial Intelligence and the Problem of Control. London: Penguin Books, 2019.
• Russell, S. and Norvig, P. Artificial Intelligence: A Modern Approach. 4th edn. Harlow: Pearson, 2021.
• Shannon, C. E. ‘A Mathematical Theory of Communication’, Bell System Technical Journal, 27 (1948), pp. 379-423 and 623-656.
• Turing, A. M. ‘Computing Machinery and Intelligence’, Mind, 59 (1950), pp. 433-460.
• Wiener, N. Cybernetics: Or Control and Communication in the Animal and the Machine. Cambridge, MA: MIT Press, 1948.