AI and Modern myths
The essence and structure of AI myths
The four interconnected AI myths
Unlimited exponential growth
Superintelligence
Computer consciousness
The singularity
Numerous influential figures express concerns about the implications of artificial intelligence. A lot of discussions have been dedicated to exploring potential outcomes, ranging from intentional to indifferent scenarios leading to humanity's elimination.On the other hand, some proponents view AI as the pivotal force capable of surpassing the current constraints of human existence. A significant theme emerging from interviews is the necessity to debunk these perceptions labelled as "myths."
It is important to remember that in this context, the term "myth" is used not to dismiss the possibility of these scenarios but to emphasize their roots in aspirations and fears, whether personal or cultural, rather than scientific evidence.
The essence and structure of the AI myth
The first level of myths pertaining to AI can be designated as the level of
clichés. Within this category, we encounter various portrayals deeply rooted in popular culture but lacking practical relevance in everyday life. These myths exist purely in the realm of fiction. For instance, there is the widespread belief that the emergence of AI will inevitably lead to a looming threat to humanity, resulting in the domination or eradication of humans by intelligent machines. Although this narrative has become a commonplace cliché, it does not evoke daily apprehension or hinder our widespread adoption of new technologies. Essentially, the clichés surrounding AI consist of fantastical depictions that exert minimal influence on our daily routines, akin to "bedtime stories" told to children. Furthermore, this category encompasses myths stemming from a misguided comprehension of AI technology's operation. Contrary to popular belief, there is no singular, all-encompassing AI capable of executing every imaginable task. Instead, the technologies utilized in AI, such as Machine Learning, Deep Learning, genetic algorithms, fuzzy logic, and inference rule-based systems, are inherently distinct entities that often overlap in the collective imagination.
While the initial level presents a certain level of clarity, the subsequent two levels exhibit intricate complexities that necessitate a comprehensive approach. Moving on to the next level, we will examine the scientific aspect of the myth surrounding AI. Comparable to the cliché level, the scientific tier undergoes systematic enhancement, this time orchestrated by specific experts or influential figures, such as Elon Musk, who promote the inevitability of this technological emergence. This designation as "scientific" is attributed to the remarkable achievements of science and technology, which not only foster optimism regarding the future of AI but also introduce a mythological element, despite the existing gaps in ongoing research. The truth remains that recent inquiries into intelligence have revealed mysteries that defy the understanding of modern science. A significant discovery concerning intelligence is the profound disparity between human-like intelligence and the current definition of AI. Nevertheless, the myth seeks to convince us that these disparities are merely temporary, destined to be alleviated or eradicated with the emergence of more sophisticated and potent technologies. For example, Ray Kurzweil predicts that by the end of the third decade of the twenty-first century, AI will achieve parity with, if not surpass, general human intelligence in all domains. However, at the time of this writing, a significant span of 7-8 years remains before Kurzweil's prediction reaches fruition, and the envisioned level is far from being realized.
While the cliché level may be largely harmless, as it does not harm the perpetuators of myths or the environment in which they propagate, the scientific level brings about detrimental consequences through its perpetuation of the myth. Ironically, at the scientific level, the mythology of AI has not faced ridicule but rather appears to be experiencing a period of flourishing. Consequently, the mythology crafted at the scientific level proves challenging to discern, as it conceals itself within a specialized vocabulary, sophisticated methodologies, theories, and technologies that elude the understanding of the average user, thereby perpetuating patterns that are detrimental to genuine scientific endeavors.
The final aspect that requires consideration is the religious dimension of the narrative surrounding AI. In the aforementioned instances, the term "narrative" was utilized conventionally to denote fiction, allegory, or a departure from reality. However, as elucidated by Alexei Losev (2008), within the realm of mythical consciousness, the narrative constitutes the most tangible and potent reality, surpassing mere fabrications or stories.
In this context, the narrative assumes an unequivocally essential role for contemplation and existence. Analogous to traditional religious ideologies, where beliefs, doctrines, traditions, and worship are unquestionably perceived as true and indispensable, proponents of the AI mythology incorporate a set of beliefs and doctrines as integral elements of a framework that seeks to elucidate reality. Furthermore, as previously highlighted, the narrative does not diminish over time; rather, it endures and transforms while retaining its fundamental purpose and structure. Consequently, we observe the emergence of new forms of secular religions.
While the myths of AI at the level of clichés have minimal social consequences, and at the scientific level exert a limited influence on the progression of scientific practices, at the religious level, the myths wield a significant impact on society.
Within the realm of the religious dimension, the myth of AI has a profound influence. It operates not solely within the technological domain but permeates the very essence of our being. It imparts narratives that elucidate our origins, shapes our perception of identity, and presents a vision of our future. By embracing the beliefs and doctrines of this myth, proponents invest AI with a transcendent quality akin to religious veneration. It becomes a focal point for belief systems and a repository for aspirations and dreams.
Thus, on a religious level, the narrative of Artificial Intelligence plays a significant role in molding societal dynamics. It intertwines technology and spirituality, giving birth to a novel form of secular belief system. By attributing personalities to abstract concepts, granting them supernatural abilities, and redefining human existence, this narrative wields a profound influence that transcends mere technological advancements to impact the very essence of human consciousness and aspiration.
Myth functions as an extensive magical construct, uniquely accessible to the mythical mindset, guiding the unfolding of historical events. It is the myth, rather than political or religious institutions, that shapes and delineates history itself. However, the myth of AI represents a reversal of historical and sacred time, projecting them into a future devoid of sacredness and sanctity. The paradox lies in the realization that once the myth materializes as tangible reality, it ultimately fades away as the mythical mindset is replaced by a "collective digital consciousness," where the myth loses its essence, identity, and miraculous nature.
In summary, it becomes evident that all layers of myths surrounding AI must be viewed as interconnected components, each complementing the others, as they serve as sources of mutual inspiration. In the entirety of human history, there has never been a time closer than the present to achieving a form of intelligence comparable to our own within a machine. Despite persistent uncertainties and limitations, the myth persists as the driving force propelling us toward this goal. The emergence of the myth of sentient machines and superintelligence was inevitable in the context of our hyper-technological civilization. Whether embraced or challenged, this myth assumes a fundamental role in shaping our modern reality.
Majority of critical discussions on the topic underline that AI, in isolation, is not destined to bring about either the ultimate demise or salvation of humanity. The examination of four interconnected myths follows: unlimited exponential growth, superintelligence, computer consciousness, and the singularity. Notably, these myths unfold in a sequential manner, with each one building upon the assumptions of its predecessor(s) andgarner substantial attention, overshadowing and impeding constructive dialogue, especially at the popular level.
Unlimited exponential growth: ‘technology is improving at an ever-increasing rate and nothing will stop it’
The notion of unlimited exponential growth often arises in discussions surrounding AI, sparking concerns about the seemingly boundless advancement of computer capabilities. This concept is frequently linked with Moore's Law, which predicts that computing power will double approximately every two years. However, a critical flaw in the conventional interpretation of "exponential growth" lies in its assumption of inevitable and limitless progress. This presumption overlooks instances of exponential growth observed only within specific time frames, such as certain chemical reactions or bacterial proliferation. In reality, beyond the realm of pure mathematics, sustained exponential growth is non-existent. Eventually, the resources fueling the growth become depleted.
Although global population growth is sometimes cited as an example of exponential growth, it's important to recognize that various factors suggest that global population growth is likely to decelerate and reach a peak of around 11 billion by the end of this century. This underscores the necessity of acknowledging the finite nature of exponential growth and avoiding oversimplified estimations of its implications, particularly within the context of artificial intelligence.
Readers familiar with the concept of exponential computer growth may argue that the historical trajectory of computing power itself serves as evidence of unbounded exponential growth. However, in reality, despite the reliability of Moore's Law over the past several decades, most experts predict its breakdown in the coming years—a view even shared by Gordon Moore himself. The imminent challenge lies in the inability to create functional transistors on computer chips that are smaller. As transistors reach diminutive sizes, they encounter quantum tunnelling, disrupting the expected behavior of the electrons they conduct. Essentially, the immutable laws of physics supersede Moore's 'Law' of computing power.
While some defend the myth of exponential computer growth by pointing out alternative avenues for progress, such as increased software efficiency, hardware specialization (e.g., three-dimensional silicon circuits), cloud computing, and quantum computing, it remains crucial to recognize that this progress is neither inevitable nor likely to be truly exponential. It is acknowledged that computing development may continue at a surprisingly rapid pace, but the emphasis should be on acknowledging the real, physical limitations of our world, including the escalating energy consumption of powerful computers. Additionally, advancements in computing are the outcome of commendable and persistent efforts by real individuals rather than being dictated by inevitable laws of nature.
The persistence of the myth of exponential computer growth may be attributed to cultural beliefs surrounding progress, which have experienced a resurgence, particularly in the unparalleled era of peace following the upheavals that began in 1914. Notable advancements in areas such as transportation, food security, life expectancy, literacy, leisure time, and medicine undeniably signify significant progress. However, it is crucial to emphasize that these technological advances are the outcomes of the dedicated efforts of real individuals, rather than being governed by inevitable laws of nature. This distinction underscores the importance of acknowledging the role of human agency and hard work in driving progress, challenging any assumption of automatic or inexorable advancement.
While celebrating technological advances, it is essential to acknowledge their darker counterparts, which are often overlooked. These include the impermanence of societal structures, surging rates of obesity, heightened levels of elderly loneliness, the proliferation of fake news, an epidemic of apathy and boredom, and unprecedented levels of mental illness. Additionally, factors like crippling levels of indebtedness and extensive loss of biodiversity cast shadows on the narrative of unbridled progress. One can argue that certain assumptions underlying AI are influenced by problematic and widely embraced interpretations of Neo-Darwinism, a perspective that itself has faced significant scrutiny in recent times. Ultimately, the concept of unlimited exponential growth, and the associated fear of 'runaway AI,' is demystified as a myth because it amalgamates scientific observations with unfounded assumptions and predictions. This highlights the need for a nuanced understanding that considers both the positive and negative aspects of technological advancements.
Superintelligence: ‘computers will soon be far better at doing everything humans do’
The present state of AI is characterized as 'narrow' due to its proficiency in specific tasks. However, there is a pursuit of Artificial General Intelligence (AGI) to expand AI's capabilities across all human tasks. Beyond AGI lies the concept of Superintelligence, which suggests that AI will surpass humans in every task. Nick Bostrom defines Superintelligence as "an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom, and social skills." This vision entails the creation of machines capable of designing and constructing even more intelligent entities.
The apprehensions surrounding Superintelligence stem from the idea that consciousness is not a prerequisite for it to pose a threat to humanity. Merely having misaligned goals could lead Superintelligence to cause harm or eliminate humans in pursuit of its objectives. However, a flaw in the theory of Superintelligence lies in its reduction of complex human activities to mere 'tasks.' This reduction partly arises from viewing humans as 'machines' and emphasizing intellect over soul and body. This perspective overlooks the essence of humanity, where qualities like love, hope, inspiration, and striving are integral and extend beyond tasks. Moreover, the notion of a 'general' human intelligence oversimplifies the diverse range of human personality and activities. While it's conceivable to envision AI excelling in various board games, the idea of a single AI-powered entity surpassing humans in every aspect neglects the intricacies that make human experiences unique.
Imagining a system that can outperform humans in every known sport or replicate every human action with equal proficiency is highly improbable. While AI may excel in specific domains, the concept of a single system matching human capabilities across the board is considered unlikely.
A critical flaw in the concept of Superintelligence is its underestimation and misunderstanding of human intelligence. There's an assumption that human knowledge represents a unique tipping point, beyond which machines can significantly surpass human intelligence. However, this assumption lacks empirical evidence, and it's uncertain whether human intelligence represents an absolute upper limit. Even if machines achieve a level of general intelligence comparable to humans, it does not ensure their ability to create new, more advanced machines.
Illustrating this point involves considering human knowledge. While a proficient professor can impart their knowledge on a particular subject to a student, this doesn't guarantee the student's ability to effectively teach that knowledge to someone else. Teaching involves higher cognitive processes than mere knowledge replication, requiring skills like communication, patience, and the establishment of sustained relationships. Creativity, as Benjamin Bloom and others have argued, is an even more advanced cognitive process than teaching. Therefore, achieving Superintelligence, as often conceptualized, involves challenges beyond replicating existing knowledge and demands a level of cognitive complexity that extends beyond mere imitation.
Despite the critiques of the concept of Superintelligence, advocates persist in their arguments, often relying on the notion of unlimited exponential growth to assert that computers will inevitably surpass human intelligence. However, the inherent weaknesses in the Superintelligence concept, coupled with its dependence on another flawed notion, lead to the conclusion that this theory is more akin to myth than science. The reliance on speculative assumptions and the overestimation of the potential for limitless growth contribute to the mythic nature of the concept, emphasizing the need for a more realistic and evidence-based understanding of the trajectory of artificial intelligence.
Computer consciousness: ‘computer systems will eventually be self-aware’
Computer consciousness, a concept initiated by John Searle's 'Chinese Room' hypothesis in 1980, has ignited extensive debate within the realm of artificial intelligence (AI). Presently, no AI system exhibits signs of consciousness, and doubts persist regarding the likelihood of this changing in the future. Many experts view the pursuit of computer consciousness as a diversion from more tangible advancements in AI.
Linked closely with myths of Superintelligence and Unlimited Exponential Growth, the idea of computer consciousness suggests that once machines surpass human intelligence significantly, achieving consciousness becomes inevitable. However, skepticism surrounding this notion underscores the need to differentiate between advanced cognitive abilities in machines and the intricate phenomenon of consciousness.
Defining consciousness itself poses a significant challenge, with divergent views on whether it entails sentience and emotions akin to human experiences. While some assert that consciousness necessitates feelings, others propose that computers could potentially be conscious without emotional experiences. Despite casual references to AI systems as 'conscious' or 'aware,' particularly in fields like Computer Vision, this usage differs markedly from human consciousness.
While AI continues to improve in emulating consciousness, mimicking consciousness does not equate to possessing it. The complexity of human consciousness transcends outward behavioral replication, presenting a challenge in defining and understanding consciousness beyond mere emulation.
The discourse surrounding computer consciousness often assumes the human brain as the locus of consciousness, suggesting replicability through methods like Whole Brain Emulation. However, comparisons between neurons and transistors overlook the intricate complexities of neural transmissions, surpassing the binary function of transistors.
Metaphorical language further complicates conceptions of computer consciousness, ascribing human-like qualities to AI systems without a genuine understanding of consciousness. Despite AI's progress in areas like Natural Language Processing, it lacks the experiential understanding inherent in human consciousness.
Language exemplifies the disparity between AI comprehension and human understanding, showcasing AI's proficiency in mimicking certain aspects of conversation while falling short of genuine understanding. The game of GO demonstrates the limited and non-adaptable nature of AI 'understanding,' highlighting the distinction between AI's simulated capacities and human consciousness.
In summary, while AI systems may excel in replicating certain cognitive functions, they remain distinct from genuine human consciousness due to the nuanced, adaptable, and experiential nature of human cognition. Thus, achieving computer consciousness remains a distant prospect, requiring a deeper understanding of consciousness itself and its divergence from AI capabilities.
Singularity: ‘a point will come when humans are no longer dominant; we will be to computers what ants are to us now’
The concept of the Singularity is among the most speculative ideas in discussions about artificial intelligence, blending assumptions from previous myths. It revolves around the idea of machines enhancing their capabilities through self-improvement to a degree where human comprehension becomes futile. This notion gives rise to various scenarios, often characterized by runaway, apocalyptic, or snowball effects.
However, belief in the Singularity faces significant challenges, particularly concerning the concepts of speed and self-improvement. Firstly, increases in computing speed or power may not necessarily lead to optimal solutions due to inherent limitations. For instance, certain problems may already have unattainable solutions, and chance elements may prevent perfect predictions despite vast computing power. This perspective emphasizes that computing power alone cannot solve all problems, as noted by Steven Pinker. Moreover, the belief in the Singularity often overlooks the complexities and unpredictabilities inherent in advanced computing systems.
Regarding self-improvement, proponents of the Singularity often assume that AI's ability to enhance itself will follow an exponential or linear trajectory. However, a more nuanced understanding suggests that AI's self-improvement may exhibit diminishing returns or logarithmic growth. Historical progress in AI development has largely followed a logarithmic trajectory, and advancing technology becomes increasingly challenging as sophistication increases.
Despite these challenges and a more nuanced understanding of AI's limitations, belief in the Singularity persists. This persistence may be fueled by the perception of an imminent cataclysmic event or the 'AI Effect,' which suggests that AI is defined by what has not yet been achieved. Some invoke Roy Amara's remark, suggesting that short-term impacts are overestimated while long-term effects are underestimated. However, it's essential to balance speculation about the future with addressing present and near-future challenges.
It's crucial to recognize that discussions about AI are often influenced by agendas beyond pure technological development. Prominent figures in AI, such as Bostrom, may have ulterior motives, such as advocacy for transhumanism. Bostrom's involvement in shaping AI discourse and policy underscores the intersection of personal beliefs and broader ideological agendas. Understanding these motivations is essential for fostering a transparent and objective understanding of AI discussions, separating scientific exploration from advocacy-driven agendas.
Conclusion
The discourse surrounding AI is rife with concerns and speculation about its potential implications for humanity. While some view AI as a transformative force with the capacity to surpass human limitations, others express apprehension about scenarios ranging from humanity's elimination to its salvation. Central to these discussions is the notion of "myths" surrounding AI, which are perceptions rooted in aspirations and fears rather than scientific evidence.
At the heart of AI myths lies a hierarchical structure consisting of three tiers: clichés, scientific, and religious dimensions. The cliché level encompasses fantastical depictions of AI prevalent in popular culture, such as the notion of intelligent machines threatening human dominance. These clichés, akin to bedtime stories, have minimal influence on daily life. However, they also include misconceptions about AI technology's operation, such as the belief in a singular, all-encompassing AI.
Moving to the scientific tier, myths surrounding AI are perpetuated by specific experts or influential figures who promote the inevitability of AI's emergence. This tier incorporates optimistic projections about AI's future based on scientific achievements but often overlooks existing gaps in research. For example, predictions about achieving superintelligence fail to account for the complexity of human intelligence and the challenges in replicating it.
The religious dimension of AI myths treats the narrative surrounding AI as a tangible reality, akin to traditional religious ideologies. Proponents of the AI myth incorporate beliefs and doctrines into a framework that shapes perceptions of reality and human existence. This narrative intertwines technology and spirituality, giving rise to a novel form of secular belief system that influences societal dynamics profoundly.
Among the interconnected myths surrounding AI, unlimited exponential growth is a prominent theme. This myth posits that technology will continue to improve at an ever-increasing rate without any constraints. However, a nuanced understanding reveals the finite nature of exponential growth and the real, physical limitations of technological advancements. Despite cultural beliefs in unbridled progress, technological advancements are driven by human agency and hard work rather than inevitable laws of nature.
The concept of superintelligence, where computers surpass human intelligence in every task, also features prominently in AI discourse. While AI may excel in specific domains, achieving human-like intelligence across the board is highly improbable due to the complexity of human cognition and the limitations of AI technologies.
Computer consciousness, another myth, suggests that machines will eventually become self-aware. However, replicating human consciousness involves challenges beyond mere emulation, such as understanding the essence of consciousness itself and its divergence from AI capabilities.
Finally, the singularity myth predicts a point where humans are no longer dominant, likening humans to ants in relation to computers. While belief in the singularity persists, challenges regarding computing speed, self-improvement, and the influence of ideological agendas on AI discourse necessitate a balanced and objective understanding of AI's potential trajectory.
In summary, discussions about AI myths underscore the importance of distinguishing between aspirations, fears, and scientific realities. By critically examining these myths and their underlying assumptions, we can foster a more informed and nuanced understanding of the implications of artificial intelligence for humanity.