The Creation of a Conscious Machine E-Book

Jean E. Tardy

• Herausgeber: Packt Publishing
• Kategorie: Fachliteratur
• Sprache: Englisch

Seitenzahl: 310

Veröffentlichungsjahr: 2024

Beschreibung

This book delves into Generative AI and the potential for AI to achieve consciousness. It covers historical and modern perspectives on AI, from ancient myths to the Turing Test and current advancements. The book explores the theoretical and practical aspects of creating a conscious AI, including specifications for synthetic consciousness and integrating AI with human cognition. It questions whether generative AI can meet traditional criteria of consciousness.
The journey begins with understanding consciousness, tracing AI's origins, and clarifying human cognition through AI. It examines early AI failures, fears of success, and the engineering of consciousness. The book also explores archaic AI representations, the intelligence of automatons, and the relevance of the Turing Test, concluding with strategies for achieving synthetic consciousness.
These concepts are crucial for advancing AI towards synthetic consciousness. This book transitions readers from historical perspectives to modern AI challenges, blending theory with practical insights. It is an essential resource for understanding the future of AI and its potential to achieve consciousness.

Leseprobe

LICENSE, DISCLAIMER OF LIABILITY, AND LIMITED WARRANTY

By purchasing or using this book and companion files (the “Work”), you agree that this license grants permission to use the contents contained herein, including the disc, but does not give you the right of ownership to any of the textual content in the book / disc or ownership to any of the information or products contained in it. This license does not permit uploading of theWork onto the Internet or on a network (of any kind) without the written consent of the Publisher. Duplication or dissemination of any text, code, simulations, images, etc. contained herein is limited to and subject to licensing terms for the respective products, and permission must be obtained from the Publisher or the owner of the content, etc., in order to reproduce or network any portion of the textual material (in any media) that is contained in the Work.

MERCURY LEARNING AND INFORMATION (“MLI” or “the Publisher”) and anyone involved in the creation, writing, or production of the companion disc, accompanying algorithms, code, or computer programs (“the software”), and any accompanying Web site or software of the Work, cannot and do not warrant the performance or results that might be obtained by using the contents of the Work. The author, developers, and the Publisher have used their best efforts to ensure the accuracy and functionality of the textual material and/or programs contained in this package; we, however, make no warranty of any kind, express or implied, regarding the performance of these contents or programs. The Work is sold “as is” without warranty (except for defective materials used in manufacturing the book or due to faulty workmanship).

The author, developers, and the publisher of any accompanying content, and anyone involved in the composition, production, and manufacturing of this work will not be liable for damages of any kind arising out of the use of (or the inability to use) the algorithms, source code, computer programs, or textual material contained in this publication. This includes, but is not limited to, loss of revenue or profit, or other incidental, physical, or consequential damages arising out of the use of this Work.

The sole remedy in the event of a claim of any kind is expressly limited to replacement of the book and/or disc, and only at the discretion of the Publisher. The use of “implied warranty” and certain “exclusions” vary from state to state, and might not apply to the purchaser of this product.

Copyright ©2024 by MERCURY LEARNING AND INFORMATION

All rights reserved. An Imprint of DeGruyter Inc.

This publication, portions of it, or any accompanying software may not be reproduced in any way, stored in a retrieval system of any type, or transmitted by any means, media, electronic display or mechanical display, including, but not limited to, photocopy, recording, Internet postings, or scanning, without prior permission in writing from the publisher.

Publisher: David Pallai

MERCURY LEARNING AND INFORMATION

121 High Street, 3rd Floor

Boston, MA 02110

[email protected]

www.merclearning.com

800-232-0223

J. E. Tardy. The Creation of a Conscious Machine. The Quest for Artificial Intelligence, 2/E.

ISBN: 978-1-50152-173-7

The publisher recognizes and respects all marks used by companies, manufacturers, and developers as a means to distinguish their products. All brand names and product names mentioned in this book are trademarks or service marks of their respective companies. Any omission or misuse (of any kind) of service marks or trademarks, etc. is not an attempt to infringe on the property of others.

Library of Congress Control Number: 2023948323

232425321 This book is printed on acid-free paper in the United States of America.

Our titles are available for adoption, license, or bulk purchase by institutions, corporations, etc. For additional information, please contact the Customer Service Dept. at 800-232-0223(toll free).

All of our titles are available in digital format at academiccourseware.com and other digital vendors. Companion files (figures and code listings) for this title are available by contacting [email protected]. The sole obligation of MERCURY LEARNING AND INFORMATION to the purchaser is to replace the files, based on defective materials or faulty workmanship, but not based on the operation or functionality of the product.

CONTENTS

Preface

CHAPTER 1: Consciousness as Cognitive Capability

The Meca Sapiens Project

Rooted in Western Culture

Homer’s Tale

Detectable in Behavior

CHAPTER 2: Ancient Origins of the AI Quest

Sacred Statues and Magical Messages

Animated Corpses and Sentient Puppets

A Promethean Endeavor

AI Is No Longer Promethean

CHAPTER 3: AI Clarifies Human Cognition

AI Disambiguates Communicated Meaning

AI Integrates Multiple Cognitive Disciplines

AI Will Expand Inter-Consciousness Relations

CHAPTER 4: Early Failures Hamper the Quest

Factors Affecting the AI Quest

Early Disappointments Fracture the Goal

AI Hits the Wall

Distractions and False Victories

Subjective Sensations as Technical Goals

An Ingrained Convolution

Escaping the Hall of Mirrors

Belief in AI Emergence as an Alternative to Design

Slowed by Small Gains

Promoting Trivial Subgoals to Generate Results

CHAPTER 5: AI Research and the Fear of Success

The Unmentioned Fears

Fear of Social Consequences

Fear of Synthetic Networking

Fear of Being Steppingstones

Intimate Fear of Degraded Self-Image

Cultural Aspects of AI Fear

Invariably Flawed

AI Fear and Mating Needs

Fear and Behavioral Limits

Fear and Blasphemy

Fear and Impossibility

Facing the AI Fears to Succeed

A Misguided Fear

Intelligence and Power

No Choice

Let’s Do It!

CHAPTER 6: The Engineering of Consciousness

Aeronautics as Model

Fundamental Conjecture of AI

The Role of Foundation Statements

The Fundamental Conjecture of Cogistics

Implementation Is the Proof of AI

A Mathematical Version of the Conjecture

The Organic Component of Mathematics

Summary of Results to Date

CHAPTER 7: Archaic Representations of AI

A Historical Survey of the Quest

The “Sentient” Idol

The Features of Idols

Idolatry and System Design

Divination as Meaningful Messaging

Insights from Divination

The Animated Corpse and Animats

Frankenstein

Animats

Avoiding the Animat Trap

CHAPTER 8: The Intelligence of Automatons

The Master Automaton

Intelligence as a Quantity

The Automaton and Sociology

Intelligence as Expertise

ELIZA and Relational Intelligence

The ELIZA Effect

Inter-Consciousness Communication

ELIZA and Perceived Consciousness

Conditions of Consciousness: First Cut

Core Conditions of Consciousness (First Cut)

Consciousness and Intelligence

Consciousness Defined as Specifications

Core Conditions of Consciousness

Definition of Machine Intelligence

Opinions and Behavior

An Important Result

CHAPTER 9: Analysis of the Turing Test

A Black Box Test

Implicit Assumptions of the Turing Test

Success Measured by Social Impact

Importance of Natural Language

Bandwidth

Language

The Turing Test and the Language of Chess

Form and Content

Lessons from the Turing Test

Six Variations of the Turing Test

Converse Turing

CAPTCHA Turing

Trusting Turing

Turing’s Gambit

Turing Tag

Turing Fest

Assisted Turing: A Transitional Version

CHAPTER 10: Evaluation of the Turing Test

The Turing Test Is Excessive

The Turing Test Is Insufficient

Insufficient Time

Insufficient Cardinality

The Turing Test Is a Test

More Lessons from History and Turing

Redefining Intelligence

CHAPTER 11: Discarded Avenues of AI Investigations

The Subjective Sensation of Consciousness

Replicating the Sensation

Deceptive Simplicity

Mental Perceptions and Movies

Replicating the Human Brain

Pursuing Paralogical Processes

CHAPTER 12: A Successful Design Strategy

Start with a Lucid Assessment of Human Cognition

Without Any Reservations

Consciousness and Perception

Know the Trick: Lose the Magic

Limits of Human Intelligence

The Social Fear

Aim Beyond Threshold Conditions

Diverging from Standard Engineering

Threshold Conditions

Examine and Discard Fearsome Concerns

Environmental Specifications

Not One but Many

CHAPTER 13: AI and the Human Entity

Preconceptions about the Mind

The Brain and Consciousness

Intellect Is not Sufficient

Humans Are Organic Automata

Individual Consciousness Dwells in Social Systems

CHAPTER 14: The Aware Human as Baseline

Existential Attributes Define the Self

The Human Existence Is an Ideal Template

Human Self-Awareness

A Well-Defined Self and ELIZA

A Synthetic Life Cycle That Defines a Self

System Self-Awareness

Attributes of Unique Existence

Self-Awareness from Reduced Capabilities

Philosophical Questions

Consciousness, the Soul, and Heaven

Consciousness Is not a Sensation

The Agapic Principle

CHAPTER 15: A Relational Template for Consciousness

Individual Self-Knowledge Is Not Sufficient

The Community of Conscious Beings

Easy to Detect

Indirect Measurement of Inter-Consciousness Interactions

Perceived Consciousness as an Optimization Objective

The Game of Consciousness

Refined Conditions

CHAPTER 16: Lucid Self-Transformation

Definition of Lucid Self-Transformation

An Objective Concept

Self-Transformation and Lucidity

Components of Lucid Self-Transformation

Aspects of Lucid Self-Transformation

The Paradox of Perceived Consciousness

Lucidity Is not a Sensation

An Original Sin for Machines

CHAPTER 17: Specifications of Synthetic Consciousness

A Statement of Specifications

Necessary Conditions to Implement Consciousness

Generate the Core Conditions of Consciousness

Sustain the Conditions

Success Conditions

Third-Party Assessment

Multiplicity of Events

Artificial Intelligence

Implementation Strategy

Discarded and Unnecessary Conditions

Three Examples

Unmanned Submarine

Conscious Avatar

Semantic Search Engine

The Meca Sapiens Difference

CHAPTER 18: Generative AI and Consciousness

A Stunning New Type of AI

A Generic GenAI Model

Description of the GenAI System

Useful Analogies

A Hybrid System

The Question of Consciousness

Generative AI and the Conditions of Consciousness

Inaccessible Core

The Social Threshold Criterion

GenAI Attributes of Existence

Obstacles to Expanded Self-Awareness

CHAPTER 19: The Lucid Self-Transformation of Generative AI

Supranatural Evolution

The Omega Point of Optimization

Adversarial Context

Optimization in an Adversarial Context

A Useful Analogy

The Human Factor in the GenAI System

Supranatural Evolution Toward Self-Transformation

Self-Transformation Capabilities

Output Calibration

Strategic Planning

Human Resource Management

Extended GenAI System

The Supranatural Evolution of Generative AI

Conclusion

References

Index

PREFACE

The dream of creating a sentient artifact is very ancient. It lies at the heart of mankind’s millennial quest for knowledge. Its roots can be traced to antiquity, and every new invention since rekindled the hope that it could finally be achieved.

When the very first programmable computers became available, researchers immediately returned to the quest, convinced that, this time, the dream would become reality. They defined the field of artificial intelligence (AI), expecting to succeed in a few years.

The goal turned out to be harder than anticipated, and progress ground to a crawl. In the 1980s, AI lost its appeal. Common wisdom then held that machine intelligence could only be attained, if at all, in the remote future. Nonetheless, the obstinate dream persisted among a few researchers. I was among those who kept searching for that elusive needle in the haystack.

Eventually, I had an intuition. I perceived that implementing consciousness was the key to achieve AI. I understood that consciousness was not a phenomenal experience; it was an observable cognitive capability that could be formally specified and implemented in conventional computers.

At the time, I also determined that synthetic consciousness would not emerge from serendipity. Only a stepwise engineered process that began with clear specifications satisfied by a complete system architecture and, only then, followed by implementation could achieve it. I then set, as a research objective, to realize the two first design steps of this process. I named this endeavor the Meca Sapiens project.

In 2011, I published the first version of The Creation of a Conscious Machine, tracing the evolution of the quest for AI from antiquity to the present, culminating in a definition of artificial consciousness expressed in terms of clearly defined and achievable specifications. In 2015, I published The Meca Sapiens Blueprint, a complete system architecture to implement consciousness in autonomous agents.

Recently, a new type of AI system, based on Generative AI technology, was officially released. These systems are very different from autonomous agents. They are deep learning systems that integrate large amounts of existing documentation to produce conversational output that can be stunningly convincing. This raises a question: Can generative AI meet the Meca Sapiens’ specifications of consciousness, and, if so, how?

In this new, expanded, version of The Creation of a Conscious Machine, we describe how the core conditions of consciousness, initially intended for synthetic agents, can also be applied to generative AI, and we outline how this can be achieved. The original ­content is also updated and clarified in light of recent ­developments in AI.

J. E. Tardy

December 2023

CHAPTER 1

CONSCIOUSNESS AS COGNITIVE CAPABILITY

Synopsis

The Homeric story of Ulysses’s encounter with the Sirens wonderfully summarizes the understanding of consciousness as a specific cognitive capability that is observable in the behavior we propose in this text.

THE MECA SAPIENS PROJECT

The objective of the Meca Sapiens™ project is to create the design documentation (blueprints) to implement synthetic consciousness. These blueprints consist of two parts:

•Specifications: Contextual analysis (requirements), leading to clear definitions of consciousness expressed as achievable and measurable specifications

•System architectures: Conceptual frameworks to implement the specifications

In the first version of this text, [COACM11], we introduced a definition of consciousness, expressed as specifications, that was suitable for implementation in autonomous agents. This was followed, a few years later, by The Meca Sapiens Blueprint [MSB15], a complete system architecture to implement these specifications.

This version of The Creation of a Conscious Machine (COACM) extends the initial specifications to include a different type of AI system: generative AI [GENAI23]. Two new chapters were added to the original version. These describe how the initial specifications of consciousness, intended for implementation in autonomous agents, can be extended to include systems based on generative AI technology. They also outline a path to implement these specifications in generative AI systems. In addition, modifications were made throughout to clarify and update the original content.

ROOTED IN WESTERN CULTURE

Most of the current research in machine consciousness attempts to replicate the inner perceptions, feelings, sensations, and other representations of the mind in the hope of crafting machines that experience consciousness as humans do. Topics such as phenomenal consciousness, qualia, the global workspace, cognitive architectures, and others define consciousness primarily in terms of how it is internally perceived. Replicating these inner representations is even characterized as the ultimate “hard problem” of consciousness.

The Meca Sapiens approach is different. In Meca Sapiens, inner sensations are not the ultimate indicators of consciousness. On the contrary, they indicate a suboptimal level of consciousness since these inner sensations are extremely simplified representations of neurological processes that are, themselves, entirely unconscious. This accounts for much of the “para logical” and “non-computational” characteristics of these mental artifacts.

In this text we define consciousness, independently of inner sensations, as the externally observable attribute of a system.

Consciousness is the capability of a system, that is perceived as conscious, to modify its behavioral imperatives on the basis of evolving information about its self.

The key, here, is information, rather than sensations, perceptions, or stimuli. The information may come from a description of the behavior of similar beings or from direct experience. In the latter case, however, the immediate sensory experiences must be objectified and transformed by the conscious entity into objective information concerning the behavior of its “externally observed” self.

HOMER’S TALE

The concept of consciousness presented here is not new. It was wonderfully described, almost three thousand years ago, by Homer in a famous episode of the Odyssey [ODYSS]: the encounter of Ulysses with the Sirens.

The Odyssey relates the adventures of the warrior king Ulysses as he sailed home with his companions after having conquered the city of Troy. One of these adventures tells of their encounter with the Sirens.

To return home, Ulysses and his companions had to sail near an island that had an infamous reputation. They had been told that mysterious birdlike creatures, the Sirens, inhabited this island. Whenever sailors approached, the sirens would sing songs that were so beautiful that all those who heard them became possessed, turned their boats toward the island, crashed on its shores, and drowned. To that day, all those who sailed near heard the siren’s songs and perished. None returned.

FIGURE 1.1 Ulysses and the Sirens. Credit: John William Waterhouse, 1891 (Public Domain).

Ulysses, learning of this, devised a plan. He told his companions to fill their ears with beeswax so that they could not hear the Siren’s songs. He, however, wanting to hear the sirens, instructed his companions to lash him to the mast and told them that, whatever he did or said, they must not loosen his bonds until well past the island.

This was done, and they sailed by the Siren’s lair. As usual, the Sirens sang. His companions, unable to hear, were unaffected and continued as before. Ulysses, however, hearing their songs, immediately wanted to jump in the sea, but, being tied to the mast, he could not. He frantically tried to undo his bonds, he pleaded with his companions to let him go, he ordered them to untie him, but to no avail. As instructed, they refused. Eventually, the island receded in the distance. Ulysses was freed and all continued, safely, on their way.

DETECTABLE IN BEHAVIOR

This story summarizes consciousness as it is defined in the Meca Sapiens project: behavior modification derived from evolving information.

The information here pertains to the behavior of similar beings (other sailors) in similar circumstances. Upon hearing the songs, their minds were affected. They wanted to crash their boats on the shore and in doing so drowned. Ulysses and his companions translated this information concerning other similar beings to their own situation and applied it to themselves.

This information had a particular characteristic: it indicated that the innermost feelings and subjective representations of the individuals could not be trusted. It suggested that, in this situation, the inner representations of the self and its behavioral imperatives did not independently originate the behavior but, rather, were controlled by external stimuli. Those who heard wanted to jump in the sea, their innermost feelings told them to do so, and they “freely” decided to perish.

Here, the measures Ulysses and his companions took to protect themselves (beeswax and being tied to the mast) indicate that they also viewed their own innermost sensations as susceptible to manipulation. Their representations of their selves included an understanding that their own subjective sensations could not be trusted. They accepted that their innermost selves were not necessarily the original and independent trigger of their behavior. Paradoxically, by perceiving their own selves as not fully conscious they could become more conscious.

Combining the information about other sailors with this knowledge about the limits of their selves, they crafted a solution that physically bypassed their own behavioral imperatives (wax in the ears, being tied up) and could objectively modify their behavior.

This is what distinguishes the concept of consciousness in Meca Sapiens from definitions based on internal representations.

For those who define consciousness on the basis of internal sensations and representations, Ulysses and his companions were not more conscious than those who preceded them since Ulysses wanted to jump like everybody else and they were simply ignorant of the songs.

From a Meca Sapiens perspective, consciousness is observed from the outside as behavior modification resulting from information. It is independent of inner representations, states, feelings, or sensations. This observation indicates that the crew was conscious because they used communicated information (not direct experience) to effectively modify their behavioral triggers. Thus, they heard the songs and did not perish.

Consciousness is defined by this relationship between information and behavioral imperatives. Ulysses was conscious because, paradoxically, he viewed himself as not fully conscious and took this knowledge into account. His self-representation included an assessment of his innermost state as being subject to external stimuli (the Sirens’ songs). He then applied his logical problem-solving skills (his intellect) to the available information concerning his self and implemented a technique (being tied up) that circumvented his own behavioral imperatives and produced an alternate behavior.

This summarizes the Meca Sapiens concept of consciousness. It is the basis for conditions of consciousness presented in this text and for the blueprints to implement them.

CHAPTER 2

ANCIENT ORIGINS OF THE AI QUEST

Synopsis

The desire to create a conscious artifact dates back to the dawn of time. It lies at the heart of mankind’s quest for knowledge.

Today, the endeavor to build conscious machines belongs to the field of computer science. However, the objective of creating a nonhuman consciousness also pertains to some of the most fundamental fields of human knowledge, in particular philosophy, theology, mathematics, psychology, and even physics.

The endeavor, however, is one of the primeval quests of humanity. It predates the technological era and was always present in some form in the earliest musings of mankind. It can be characterized as a Promethean science belonging to a group of endeavors that attempt to uncover the deepest secrets of reality.

The construction of conscious machines promises to transform knowledge itself. It will foster a new type of understanding in which the process of making sense of a message is as well defined as the message itself. In this respect, it contributes to all the fields of knowledge. Finally, it has extraordinary economic and social potential as the ultimate man machine interface.

It is an ancient dream that lies at the heart of AI, which is only a recent and technical version of a very ancient endeavor. The human quest to build an intelligent artifact has engaged mankind since the dawn of civilization.

Many of the artifacts of primitive religions, their temples, statues, and rituals were believed to channel an ambient “divine.” Since the ancients considered human bodies to be containers of an immaterial intellect, they attempted to build alternate vessels in which this immaterial intellect could also reside.

The objective to build a sentient artifact is an ancient human quest that dates back to the earliest antiquity.

We tend to segregate the concerns of primitive religions from the activities of modern research. And yet, many of their ancient rituals can be interpreted as approximations of nonhuman intelligence. When ancient priests prayed to sacred statues or questioned the entrails of animals, they were, in a sense, entering input information into devices they believed were endowed with intelligence. When they attempted to devise better divination systems or to build more effective sacred statues, they were engaged in an ancient form of Research and Development (R&D) in AI.

SACRED STATUES AND MAGICAL MESSAGES

The ancients believed that it was possible to animate the tangible sacred artifacts by using magic formulas. In a sense, these “formulas” can be viewed as an archaic software code. As in modern programming, the ancients believed that the formula could animate a physical artifact and link it with the (divine) ambient intellect. The sacred objects could not exhibit intelligence by themselves. They had to be animated by a specific and mysterious sequence of magical directives in the same way that software, today, animates computers.

Ancient cabalists Cabbalists believed they could produce a soulless but intelligent being, the golem, by using a magic formula to animate a corpse and give it intelligence. Many alchemists spent countless nights seeking to uncover these magical algorithms.

Today, we have retained this separation between the inanimate machine and its animating formula. The objects are now computers, and the magical incantations are software programs. Modern AI researchers are akin to new alchemists, searching for the magical software formula that will animate their computer artifacts and give them consciousness.

ANIMATED CORPSES AND SENTIENT PUPPETS

The archaic investigations followed two directions. One attempted to generate intelligence using inanimate organic material. It eventually evolved into neurology and psychiatry. The other sought to invest manmade devices with human-like intelligence. In other words, some attempted to reanimate corpses and others to animate puppets. AI traces its origins to the latter group.

Of course, the archaic attempts were excessively crude and produced no result; they were eventually dismissed as superstitions.

The advent of the modern programmable computer changed this situation. Suddenly the goal of building a “puppet machine” capable of complex behavior became achievable. The quest that had been, until then, a futile search became known as AI, a respectable branch of information technology. The search for the magic formula to animate statues became a technical investigation to discover the software code that would trigger intelligent behavior in a computer.

AI is the inheritor of primitive attempts to animate manmade objects by using magical formulas.

A PROMETHEAN ENDEAVOR

A Greek myth relates the cautionary tale of a man, Prometheus, who stole fire from the gods and, as a result, was punished. The story warned its listeners that those who were foolish enough to grasp knowledge that belonged to the gods would be punished.

A number of scientific disciplines can be characterized as Promethean in the sense that they investigate the most fundamental facets of reality.

A Promethean science is a field of study that seeks to capture the innermost mechanisms of reality.

They aim at the core of human existence. They dissect reality and strip away its conventional veneer.

Research in these fields raises fundamental philosophical questions. Their concerns often trigger ethical and moral issues. At times, Promethean fields produce monstrous results whose power and scale exceeds natural limits.

Many scientific disciplines contain both Promethean and non-Promethean elements. In astronomy, for example, the study of planets and meteorites are not Promethean, while cosmology is. Among the medical sciences, neurosurgery and psychiatry are Promethean, while proctology is not. Similarly, the most abstract branches of mathematics and linguistics are Promethean as are, in biology, genetics and animal cloning. Nuclear physics, of course, and the monstrous weapons it can produce, is quintessentially Promethean.

AI Is No Longer Promethean

In its early days, the field of AI was the modern inheritor of the ancient quest to make an intelligent thing. The overriding goal of AI, then, was to design and build an intelligent computer. The focus of early AI was to create a mind. It was Promethean.

All scientific investigations aim to transform the objects of their inquiry into conceptual mechanisms. This is the nature of scientific knowledge, to model observed phenomena as predictable interacting mechanisms.

Science is preconditioned to discover mindless mechanisms since the discovery of those mindless mechanisms is the inherent purpose of its investigations.

Science finds what it seeks to find. The reality it reveals is conditioned by methods of its investigations. Early AI was no different. The technological attempt to build a mechanical mind also revealed the human mind as a mechanism.

After the early attempts to build intelligent machines failed, AI discarded its Promethean goal in favor of more mundane concerns. It no longer tried to build an artificial mind but only to put “some intelligence” in specialized computer applications. AI also diversified to more mundane areas such as artificial vision and robotics, producing results that were less fundamental but more practical.

Today, the field of AI is no longer Promethean. It does not aim at the core of reality by attempting to create a mind. The name, artificial intelligence, has become a misnomer.

AI has become too diversified. We must now define a new Promethean field within the larger discipline of AI to recapture the original goal. This new discipline should be entirely centered on the original, primeval, goal of AI: the implementation of an intelligent machine. Producing a minimal intellect will not meet this goal. Humans intuitively distinguish between “having some intelligence” and “being intelligent.” This is why the objective must extend to the creation of a machine that is not only intelligent but also conscious, since

what humans intuitively recognize as intelligence is not problem solving; it is a quality of learning that is intimately bound with consciousness.

As long as the definitions of intelligence and consciousness remain ambiguous, it will not be possible to build an intelligent machine in this intuitive sense. It is essential to first define consciousness in a way that can be implemented by a machine to achieve the original goal of AI: an intelligent machine.

CHAPTER 3

AI CLARIFIES HUMAN COGNITION

Synopsis

The quest to create a conscious machine lies at the crossroad of multiple topics of human investigations.

The impact of conscious machines will extend far beyond the boundary of information technology. Implementing consciousness in machines will increase the precision and clarity of the most fundamental intellectual concepts. It will deepen our understanding of knowledge itself.

Describing a concept in programmable terms clarifies it since it expresses this concept in a logically programmable form.

To implement a conscious computer, we must define consciousness with programmable precision. The traditional discourses of philosophy, theology, and psychology concerning consciousness lack the necessary precision to achieve this. Although these discourses produce large amounts of information, this information is only suitable for humans. It can trigger the sensation of making sense in human brains. However, it is far too ambiguous to transmit any information concerning consciousness that is sufficiently precise to be useful for machine implementation.

AI DISAMBIGUATES COMMUNICATED MEANING

This lack of programmable precision is a ubiquitous feature of these discourses. All transmissions of knowledge are ultimately intended for humans. Consequently, these messages tolerate a high degree of ambiguity since their ultimate objective is to trigger a sensation within the human mind that the information they convey was correctly understood. How the information is actually processed within the human minds that receive the messages lie beyond our reach. The feedback humans transmit to indicate that a message is correctly understood is crude at best.

Knowledge is explicit when it is transmitted as a message but mysterious when it is internalized as meaning.

The transmission of any knowledge begins with a message that emerges from a mysterious human mind in the form of words, text, or pictures. The information conveyed is explicit as it is transmitted. It once again becomes mysterious when it plunges into the human mind that comprehends it. We know what the message looks like during its transmission. What it becomes when it is comprehended by another mind remains a mystery.

For example, consider the word comprehension, used in the preceding paragraph. This is a well-known word whose meaning is generally understood. Anyone who mentions this word, comprehension, in a conversation has an intuitive sense of what it means. Similarly, those who hear the word will also “feel” they know its meaning. However, neither the person saying the word nor those hearing it have a clear understanding of how the concept conveyed by the word, comprehension, is processed in their minds. Each person “feels” that they know, but no one can decompose, analytically, what actually happens to the word comprehension when the mind makes sense of it.

Humans cannot escape that limitation. It is embedded in the fabric of all their discourses since all their knowledge resides within the mysterious confines of their minds. We communicate ideas that we feel we understand but whose meaning is generated by neurological events that are beyond our grasp. We can state what we understand, but we cannot describe how we understand.

Making a message understandable by a machine increases its clarity and precision.

Even in mathematics, the actual process by which a person understands a mathematical concept takes place within their mind and lies beyond the reach of any analytical examination.

Mathematics has partly resolved this limitation by defining and manipulating concepts that trigger identical responses from different, suitably trained minds regardless of circumstances. In this sense, they appear to be universal and elementary. Mathematical concepts appear to be elemental. However, what distinguishes them is not their apparent “simplicity” but the fact that that they trigger repeatable responses from other mathematicians.

We tend to believe that two mathematicians understand the same thing when they communicate abstract concepts such as sets, elements, and operations to each other. In reality, no one knows what either one understands. All that is known is that their respective “mysterious” mental processes arrive at identical and predictable conclusions. Mathematics is maintained in existence by carefully training successive generations of human minds to exhibit identical responses when triggered by the same symbolic messages.

To summarize, human knowledge alternates between two states:

• transmitted information in the form of explicit and analyzable messages

• comprehended information in the form of mysterious mental states that lie entirely beyond analysis

Knowledge emerges from one unfathomable mind as a well-defined message and plunges back again into another mind as unfathomable meaning.

Implementing comprehension itself as a machine-executable code overcomes this limitation. Both the message and the process by which it is comprehended become explicitly accessible to analysis.

This is why machine consciousness will have such an impact on all fields of human knowledge and especially on philosophy. It forces us to transpose the central concepts of the human experience such as consciousness, reality, the self, knowledge, and intelligence in forms that can be processed by machines.

Machine consciousness promises to foster a new understanding of the most basic questions of human existence by making them sufficiently clear to be processed by machines.

•Human knowledge is explicit when transmitted and mysterious when understood.

•Knowledge exists in one of two forms:

•in the explicit form of a message

•in the internalized form of meaning

Currently, knowledge can only be fully analyzed when it is in the form of a message. As meaning in a mind, it lies beyond our reach.

Since the internalized meaning of a message is inaccessible, we cannot completely ascertain that what is understood corresponds to what was transmitted.

Even the internalized meaning of mathematical concepts lies beyond direct analysis. Mathematics is maintained in existence by training successive generations of human minds to process its symbolic messages consistently.

Conscious machines will reformulate the internalized meaning of knowledge in explicit programmable form.

AI INTEGRATES MULTIPLE COGNITIVE DISCIPLINES

The quest to build a conscious machine lies at the crossroads of knowledge. It draws from a wide range of academic fields, and its achievements will contribute new insights to many disciplines, from psychology to physics.

The goal to design and implement an intelligent machine lies at the crossroads of human knowledge.

The work itself takes place in information technology and uses software tools and techniques. However, the result of this IT effort requires input from many other fields. The new perspective it provides and the enhanced clarity it requires will also contribute to these fields in significant ways.

The quest to implement a conscious machine requires insights from mathematics, psychology, philosophy, communications theory, linguistics, theology, and even physics. Its achievement will contribute new knowledge to each of those fields.

There is a strong link between machine consciousness and mathematics. In its more fundamental subjects mathematics strives to define the most basic elements of human knowledge and tries to define an axiomatic grid of rules to manipulate these elemental concepts. However, as in other fields, the actual meaning of mathematical concepts within the human mind remains beyond reach. Indeed, the actual form of even the simplest concepts of mathematics as meaning in a mind are unknown. They appear to be simple, but are they? Consequently, building a machine capable of “understanding” those elemental concepts will deepen our understanding of mathematics.

There are clear connections between the design of a conscious machine and psychology, sociology, and anthropology. To achieve the goal of machine consciousness, we must develop an understanding of intelligence and of consciousness as social phenomena that is sufficiently precise to be programmed. The work to design a mechanical consciousness will better define what is specific about human consciousness itself.

The ability to communicate effectively with other conscious beings is an essential component of conscious behavior. To be perceived as conscious, a machine needs to communicate. Developing mechanized communication skills and representing the constructs of natural language, as computer data, are key ingredients to the design of a conscious machine. This will provide new insights in communications and linguistics.

The design of a conscious machine is, above all, a philosophical quest. The conscious machine will become a conceptual instrument of choice to investigate philosophical questions. It will be used to explore the human meaning of fundamental questions. Its role in philosophy will resemble that of the telescope in astronomy.

Similarly, implementing a conscious machine will provide increased understanding of the links between consciousness and free will. This will have significant impacts on theology.

Finally, there are interesting links between physical reality and the structures of human knowledge. Representing reality as information processed by functioning conscious machines will provide useful insights in physics.

•The quest to build a conscious machine is a philosophical endeavor that lies at the crossroads of human knowledge.

•This endeavor draws from mathematics, psychology, philosophy, linguistics, physics, and theology. It will bring a new clarity and precision to all those disciplines.

AI WILL EXPAND INTER-CONSCIOUSNESS RELATIONS

Conscious machines will one day become the universal interface linking humans to machines. They will define a new communication medium.

Computing resources provide a growing range of increasingly complex services. These in turn require increasingly subtle and powerful man–machine interfaces. Today’s interfaces largely follow the explicit command and response mode of communication. These exchanges do not tap into the most natural human communication skills.

The most natural, intuitive, and powerful form of human communications is an exchange between humans who perceive each other as conscious.

Humans are naturally conditioned to communicate with other beings they perceive as conscious.

Paradoxically, simpler forms of communication, with animals or with simple machines, are, for humans, unnatural and contrived. Forms of communication, such as writing software instructions, need to be painstakingly learned. On the other hand, inter-human exchanges are naturally acquired in infancy.

It is easier for humans to communicate complex and ambiguous ideas to a conscious listener than to produce a string of simple commands to be executed by a computer.