The Question Of Intelligence
The Term Artificial Intelligence
Defining Intelligence
Intelligence As A Cultural Construct
Intelligence In The Context Of Ai
Ai And Human Intelligence
The Term Artificial Intelligence
In one of our previous lectures, we discussed how the term Artificial Intelligence was coined by John McCarthy and his collaborators in 1955, to describe the use of machines to replicate some aspect of human thought. According to McCarthy, there were many alternatives such as ‘‘engineering psychology’, ‘applied epistemology’, ‘neural cybernetics’, and ‘non-numerical computing’, before AI became the definitive label.
Despite the intense interest at present, the term AI remains controversial as there is a lack of a widely accepted definition for it. This controversy and ambiguity are often traced back to McCarthy’s initial aim of coining a phrase that would be broad, and attention-grabbing, without being tied to any particular scientific field or body of knowledge. The phrase ‘AI’ also fulfilled McCarthy’s need for a name that was not linked to any particular approach to computing and the emulation of thought. The term ‘artificial’ broadly meaning, ‘made by humans’, was not limited to the many ways in which humans were at the time attempting to make computing devices, such as cybernetics or automatic programming.
Defining Intelligence
Although the term ‘artificial’ is important to the meaning of ‘AI’, it is the term ‘intelligence’ that is definitive both in setting the research agenda and evoking its enormous potential. However, the concept of ‘Intelligence’ is highly debated in meaning and full of ideological connotations.
Defining "intelligence" has been notably challenging over the centuries. Some mutually influencing issues that come up in discussions about intelligence include uncertainty regarding the components of intelligence – the question of whether it represents a general mental capability or arises from various cognitive processes, the question of cultural variations in assessing intelligence, concerning the universality of the concept of intelligence, and the delineation between intelligence and non-intelligence, addressing the role of factors that are not purely intellectual yet significantly impact intelligent behaviour.
Previously, psychologists often evaded these issues by claiming, that intelligence is what intelligence tests measure. However, most experts agree that "intelligence" refers to a very broad capacity to solve an extensive array of diverse problems. Some of the commonly acknowledged primary criteria for high human intelligence include: Acquiring a broad spectrum of new knowledge and skills, mastering language, and engaging in meaningful conversations, possessing proficient short-term memory and comprehending the purpose of actions - including those of others, employing existing knowledge and skills to learn new tasks, gaining abstract knowledge by generalizing from more concrete forms, having emotional intelligence, recognizing and appropriately responding to emotional states in oneself and others and demonstrating these abilities even with limited knowledge and time available, and so on.
It is important to note that the word “intelligence” didn’t really exist for a long time, although the concept it represents has been a topic of discussion since ancient times. Aristotle used the term “reason” to signify the capacity of humans to overcome their animal instincts, a trait he believed animals lacked. Charles Darwin, on the other hand, introduced much finer distinctions, fundamentally altering the way humans perceived themselves, through his evolutionary theory. He also wrote about “reason,” describing various levels of this attribute, observed with varying degrees among individuals. Darwin's observations in evolution led him to infer that "mental powers" were much stronger in more advanced species.
George Romanes, one of Darwin's younger acquaintances, discussed animal intelligence in his 1893 book on the mental evolution of animals. Since then, the term "intelligence" has become widely utilized to label something that remains unknown, likely not existing in the manner we perceive. Humans have the ability to transform a creation of their own mind, into a part of a shared accepted reality. Religions, social norms, currency, and even corporations, all fall under such intersubjective products as they cannot exist unless a large body of people collectively believe in their existence.
According to historian and philosopher Yuval Harari, the immense success and spread of Homo sapiens stem from their capability to generate shared myths, thereby mobilizing large groups of humans. As far as our current knowledge extends, no other species seems capable of embracing fictional narratives, not even the species Homo erectus, which existed for approximately two million years and became extinct around the time of Homo sapiens' emergence, and perhaps not even Neanderthals, whose intelligence has been discovered to surpass previous estimations. The disappearance of these two human species might be interpreted as a consequence of the activities of Homo sapiens, whose distinctive mental ability is likely not intelligence but the capacity to establish and sustain intersubjective reality.
Intelligence As A Cultural Construct
In 2007, DeepMind scientists Marcus Hutter and Shane Legg, one of the co-founders, attempted to distil the concept of intelligence by collecting existing definitions in their paper titled “Collection of Definitions of Intelligence”. The presence of numerous, greatly diverse definitions and models of intelligence indicates that intelligence is not a tangible but a conceptual entity.
Moreover, different cultures attribute different meanings to human intelligence. The generally accepted definitions of intelligence often stem from the "Western" cultural context. In numerous Eastern cultures, intelligence primarily links to the social roles and responsibilities of individuals. In Chinese culture, intelligence also refers to the capacity to empathize and understand others. Similarly, several African communities view intelligence through a social-emotional lens. Sometimes, emphasis is placed on social responsibility over specific social roles. For instance, within the Chi-Chewa language spoken by around 10 million people in Central Africa, intelligence encompasses not just cleverness but also the ability to assume responsibility. Even within Western culture, there exist diverse interpretations of intelligence.
Most importantly, culture also influences individual thinking, so the culture or subculture of an individual will influence the individual’s notion of intelligence. In other words, the process through which personal perspectives become shared among a group is not a unilateral process but a dynamic interplay of mutual influence and exchange.
Until the twentieth century, the concept of intelligence had largely remained on the periphery of English-language discourse, which is surprising when considering its current important role in numerous discussions, not limited to AI. What also astonishes contemporary users of the term is its historical link to Eugenics, and ideologies of white supremacy, colonialism, classism, and patriarchy.
The notion that the most intellectually capable individuals should govern others has ancient roots, tracing back to Ancient Greece. However, its broad acceptance in the Western world emerged when it became a supporting ideology for colonialism. Aristotle's argument, asserting that certain individuals, due to their superior intellect, were destined to rule while others were meant to be ruled, provided justification for the subjugation and enslavement of non-Western populations. The concept of race underpinned this narrative, depicting non-white races as intellectually inferior—echoing Rudyard Kipling's description of them as 'Half-devil and half-child'—which, in turn, justified the perceived right or obligation of white people to dominate them and their territories.
During the nineteenth century, the Western scientific establishment actively attempted to provide evidence for these claims in a more systematic manner. English scientist, Sir Francis Galton following a series of tests measuring intellectual ability, coined the term 'eugenics.' He defined it as 'the science of improving stock' with the aim of giving more favourable races or bloodlines a better chance of prevailing quickly over less favourable ones. This concept intertwined with the global spread of the eugenics agenda, where intelligence became the pivotal factor in determining which populations should thrive and which were considered 'less suitable.'
Intelligence, especially through intelligence testing, played a significant and intricate role in the twentieth century's political history. It wasn't merely used to justify racist and imperialist ideologies but also to endorse patriarchy and classism. While the most severe instances of exploiting individuals considered less intelligent were prevalent in the first half of the century, intelligence testing has maintained its importance to date. Simultaneously, associations between varying levels of innate intelligence and different races, genders, and classes persist.
There is good reason to assume that these associations likely influence the expectations for AI. For instance, they could fuel worries about job displacement among middle-class white professionals like lawyers or doctors, whereas it's the disadvantaged and marginalized communities who are at a higher risk of negative impact. Additionally, linking this technology with the term "intelligence," which is historically associated with white men, might hinder opportunities for women and people of colour in this crucial sector, perpetuating an unjust cycle.
Intelligence In The Context Of Ai
So, what does ‘intelligence’ imply when it comes to AI? Why is AI possible? How is it related to human brain intelligence?
In science fiction, AI often takes on the guise of an extraordinary, sometimes boundlessly expansive mental force. However, in reality, AI functions as a far more commonplace entity. Surprisingly, many individuals might not realize that their initial interaction upon waking in the morning involves AI. A smartphone, aside from performing basic tasks like setting alarms, serves as a compact reservoir of AI, nearly always carried along. The AI within has a significant amount of computational power at its disposal.
When it comes to AI, it is extremely difficult to develop the mental functions that even an individual with an intellectual disability or a two-year-old can easily produce. These functions, honed through millions of years of natural evolution, are intricate processes. Consider the simple act of reaching for an object—an action demanding our nervous system to orchestrate a sequence of physical commands, three-dimensional perception, and feedback. It seems like an easy task because the software in our brains is designed to be capable of performing it. However, replicating such tasks in robots remains a challenge.
Similarly, a machine may have more precise perception, learn better than a human, and find solutions better than and different from human thinking, but the adaptive success of the human brain does not lie in simply solving the situation at hand. Rather, it thrives due to its adaptability, specifically its capacity to learn directly from the cerebral activity related to the situation. The brain doesn't merely gather outcomes of mental activity for learning; it also records the process leading to these outcomes, incorporating this path into its toolkit of skills.
Moreover, the human brain can imagine things that don't exist and create a cohesive whole from almost anything. To equip AI with human-like abilities, it needs a dual function at the perception level: a capacity to perceive the essence from the general and identify or understand unseen elements as a unified whole. This means being able to understand the most important parts from the bigger picture and to recognize or explain new things as a whole, even if they're odd or don't fit with what's known. Along with having a lot of information, the AI also needs to use imagination to understand even really unlikely combinations. Right now, machine learning is just beginning to include these kinds of processes. Without a specific way to do this, even the most powerful AIs won't be able to solve simple but unexpected problems.
Ai And Human Intelligence
The development of AI relies on varying degrees of research in brain science. The operation's enigmatic "black box" characteristics closely relate to brain science's incomplete understanding of the rules and mechanisms governing human brain intelligence. Unveiling AI's "black box" depends on further advancements in brain science.
Is it plausible for AI to surpass human biological intelligence in the future? Such predictions assume significant potential for machine intelligence but overlook the reality that human brain intelligence remains underutilized and holds substantial potential. Concerning AI, it covers only a fraction of human intelligence, primarily the conventional, logical, explicit, and universal consciousness. The challenge lies in simulating or expanding the unconventional, illogical, and personalized aspects of consciousness, which remain difficult.
The development of AI is dependent on research in brain science. The operation's enigmatic "black box" characteristics closely relate to brain science's incomplete understanding of the rules and mechanisms governing human brain intelligence. Unveiling AI's "black box" depends on further advancements in brain science.
Is it then possible for AI to surpass human biological intelligence in the future? Such predictions assume significant potential for machine intelligence but it also overlooks the reality that human brain intelligence remains underutilized and holds substantial potential. Concerning AI, it covers only a fraction of human intelligence, primarily the conventional, logical, explicit, and universal consciousness. The challenge lies in simulating or expanding the unconventional, illogical, and personalized aspects of consciousness, which remain difficult.
In conclusion, AI is not independent or self-sufficient; it falls within the scope of human intelligence, constituting a segment of it. The reason why AI belongs to human intelligence is that it is a product of human intelligence developed to a certain historical stage, a tool of the human brain, and an expansion of human intelligence. All forms of AI evolve in alignment with human intelligence, operating within the bounds of human reflection and control across various societal domains. The increasing complexity of contemporary social life poses challenges for direct adjustments and control by the human brain. As a result, fulfilling societal demands both in scale and precision becomes progressively difficult for humanity. This isn't a limitation of human intelligence in quality but rather in quantity. Hence, expanding the brain's capacity presents a possible avenue for advancing human intelligence to meet the demands of social life.