Artificial Intelligence and Consciousness: A Psychodynamic Approach
This treatise is motivated by the observed superficiality and restrictive frameworks that characterize much of the discourse surrounding Artificial Intelligence and consciousness.
This superficiality is evident across the spectrum, from expert practitioners to leading figures in the field. A fundamental cause of this shallowness is the manifestation of the discussants' limited understanding of human cognition or, at best, the neglect of psychology, the discipline that endeavors to explain humanity in human terms. The reasons for this omission are a matter of significant inquiry, potentially warranting further investigation. Whether this reflects self-alienation, a conflict of interest, or self-censorship remains to be determined.
The crux of the issue lies in the exclusive cognitive framework adopted in these discussions. Within this paradigm, intelligence is treated as a purely cognitive phenomenon, proceeding on the reductive assumption that its artificial counterpart is shaped by the same dynamics as that possessed by humans. From this juncture, it is extrapolated that consciousness will naturally emerge after a certain threshold is reached. The validity of this expectation of natural progression, and the plausibility of a cognitive phenomenon transitioning to consciousness, remain in question.
While these inquiries are comprehensible, the constricted domain within which they are posed does not provide a conducive environment for their resolution. This flawed approach inevitably precipitates either preconceived notions or an entrenchment in intractable philosophical debates.
For instance, Nobel laureate mathematician Geoffrey Hinton, regarded as a progenitor of artificial intelligence, recently disassociated himself from artificial intelligence research. He attributed his departure to his apprehension regarding the attainment of consciousness by artificial intelligence. In numerous interviews, he has sought to substantiate the fundamental causes of his apprehension through cognitive explanations.
Another perspective is offered by Nobel laureate physicist Roger Penrose. He posits that artificial intelligence can never attain consciousness because its connection with the physical world cannot reach the level of human interaction. According to Penrose, this cannot develop through chips. For him, consciousness is a cosmic phenomenon rather than a mathematical one. We concur with this view. However, Penrose's substantiation involves mathematical and astrophysical explanations that are virtually incomprehensible to the general audience. While his conviction is evident as he explicates his approach, the strain he experiences in conveying it to his interlocutor is equally apparent.
The Turing Test paradigm is another basis for evaluating the artificial intelligence-consciousness relationship. The Turing Test employs a phenomenological approach, focusing on observable behavior rather than the intrinsic nature of consciousness. In this framework, successful communication between an entity and a human, without discernible differentiation, signifies a successful test. While pragmatically useful, this approach has limitations.
When relating language-communication to consciousness, many things are overlooked. For instance, the loss of speaking ability does not equate to the loss of consciousness. Furthermore, the connection established between language and consciousness is not absolute. We can say there is a relationship between language and consciousness. However, evolutionarily, consciousness must have developed before language. Otherwise, it would imply that all humans who lived before the development of language, comprising the longest period of human history, did not possess consciousness. There might have been a leap in consciousness, but this must have occurred before language. Consciousness can very well exist without language. We can even observe this to a certain degree in other species. The natural consequence of all these explanations is that communication through language cannot prove an entity's consciousness. Being convinced as a result of communication is one thing, and that entity actually being conscious is another. Even without conversational exchange, interactions with living beings such as chickens, cats, and dogs often evoke a sense of their consciousness. In conclusion, establishing an inseparable relationship between language and consciousness, and testing the existence of consciousness through language, is both inadequate and contradicts our observations.
Now, let's step outside the confined framework of these debates and consider how we might approach the subject from the perspective of depth psychology or the psychodynamic approach.
Let's consider the period before a baby starts speaking in their development. It is important to remember that a child's development, in a way, reflects sections of the entire evolution of humanity. A child gains their sense of self-awareness long before language development. They cannot express this in this interim period, but many observations have experimentally shown the child's awareness of themselves. The bond between language and consciousness is an organic one, and consciousness develops first.
Let's look at these developmental stages: While babies make sounds like 'goo' in the first months, they begin to babble with syllables like 'ba-ba' around 4-6 months, and usually say their first words between 12-18 months. Subsequently, between 18-24 months, their vocabulary rapidly expands, and they begin to form simple sentences. By 2-3 years old, they can form more complex and understandable sentences. If language develops before consciousness, at best, babies cannot be conscious until one to one and a half years old, and at worst, until three years old! To pass the Turing test, they need to grow up to 6-7 years old! Therefore, the observed developmental stages of infants support the assertion that consciousness precedes language, challenging the notion that AI could achieve consciousness solely through linguistic proficiency.
From all of this, it is possible to draw the following conclusion: Consciousness comes first. Language may or may not emerge after consciousness. We see this in autism. Language development in autistic individuals can be quite limited. However, we cannot claim that they are not conscious. They perceive their environment and can sometimes develop exceptionally advanced relationships.
Now, if we ask the question the other way around, can consciousness emerge from language? Can there be an egg without a chicken? That is, expecting a being called pure language, whatever that means, to produce consciousness when consciousness has not existed... It is understood that Artificial Intelligence is currently developing through Large Language Models (LLMs). The fundamental modeling in this development is based on neural network modeling. This model is designed inspired by the structure of the human brain. It has the feature of self-improvement, and while doing so, it obtains language usage from the internet's large database and reshapes itself again and again with user feedback. But the basic foundation is mathematics and language. However, let's not forget that consciousness must develop before both mathematics and language. For this data, method, and design to develop consciousness on its own requires us to accept consciousness as a phenomenon that can emerge independently of the organism, the human, somehow here and there. In such a case, what we are talking about means something different from human consciousness as we understand it. So, those who don't believe in ghosts are now trying to convince us that ghosts can exist.
Since consciousness is a human attribute, let's now address what it means to be truly conscious from a psychodynamic perspective, compare it with what is attributed to Artificial Intelligence, and see how close Artificial Intelligence has come and can come to this.
The psychodynamic approach, fundamentally still influential, is development psychology-oriented, primarily Freudian but not limited to it. Within this framework, consciousness is not treated as a static phenomenon existing on its own. The consciousness mentioned here exhibits an organic and dynamic existence. According to this approach, we cannot think of consciousness as a phenomenon independent of humans. It is a phenomenon within dynamic power relations and conflicts. Because the being that carries it, namely the human, is an integral part of nature, which both creates and nourishes them but also threatens them. Consciousness is at the center of these conflicts. A consciousness without conflict, if it exists, is only a cosmic consciousness. This is the state that Buddhists or Sufi dervishes try to reach and sometimes claim to have reached. This should be noted, as there are those who claim that Artificial Intelligence can reach this stage one day.
When attributing a phenomenon like human consciousness as we know it to another being, loading it as a static ability without considering the dynamism it contains is a very simplistic and incomplete approach. We know that AI progresses through neural models. The potential of these models to produce consciousness in humans is the most attractive and current interpretation. Let's continue from the psychodynamic approach, especially avoiding making philosophy.
This consciousness has a Below, an Above, and a Middle. There must be a subconscious (id) that constantly expects satisfaction of fantastic desires and creates pressure, a Super-ego that forces compliance with external and internalized norms, and a central Ego that tries to regulate the conflict between these two dynamics. In other words, consciousness is an extremely dynamic phenomenon with a below, a middle, and an above. Consciousness is not merely an entity that knows, comprehends, understands, and enables self-awareness. If we are talking about the consciousness we think belongs to humans, then the being possessing it must also reflect the psychopathological conditions we frequently perceive. Why is this so? Because the human is the product of a challenging journey. With everything. And these challenges, while creating the consciousness it possesses, also test it in many different ways. When these tests reach points it cannot handle, consciousness tries to resolve them in some way. One way or another, it develops certain defense mechanisms. No conscious human on earth can completely escape this conflict. When defense mechanisms are also strained, which happens very often, psychopathological conditions arise.
The price of being conscious for humans encompasses psychopathological conditions. For example, anxiety disorders, feelings of helplessness and hopelessness, depression, paranoia, suicidal and self-destructive thoughts and behaviors, obsessions, compulsions, delusions, hallucinations, and even psychotic attacks may manifest. These conditions are sometimes the fundamental sources of creative dynamism. Being conscious also entails confronting the reality of mortality, both one's own and that of others. The consciousness possessed by the human being is of this nature. It is not a purely rational entity that knows everything, calculates flawlessly, and consists solely of language and intelligence. Such a consciousness has never existed.
The human is a product of this universe. The age of the universe is approximately 13 billion years. The differentiation of humans from other species on Earth spanned approximately 6 million years. Its embodiment into the form we know occurred 300,000 years ago. One must visualize the density and magnitude of data that facilitated the evolution of the human being. Were one to attempt to upload this data to servers, it might necessitate the utilization of the entire Milky Way galaxy as a server infrastructure. Only then might one approximate what we refer to as human consciousness. In this context, Penrose's fundamental assertion holds merit. However, the methodology he employs to elucidate this assertion is problematic.
We maintain that Artificial Intelligence cannot attain consciousness. The impossibility of this is not merely technological but fundamentally cosmological. No entity within the universe, or within the scope of our discussion, can transcend cosmological limitations. Claims or convincing demonstrations of such transcendence do not validate its occurrence. Human consciousness imposes a burden that is excessively heavy for any other entity to bear. Even humans continue to grapple with this burden.