This very question caused a former Google employee named Blake Lemoine to lose both his job and his reputation. You have probably read about it.
In 2022, while testing one of Google's AI systems called Lamda, Blake felt that the AI was conscious. He did not stop there. Instead, he began advocating for the rights of that AI. That eventually cost him his job.
Before talking about the consciousness of AI systems, let us first understand our own conscious experience.
In 2022, while testing one of Google's AI systems called Lamda, Blake felt that the AI was conscious. He did not stop there. Instead, he began advocating for the rights of that AI. That eventually cost him his job.
Before talking about the consciousness of AI systems, let us first understand our own conscious experience.
How we acquire conscious experience has long been a puzzle that troubled neuroscientists. When modern tools such as functional MRI scanners were invented, neuroscientists became capable of explaining various perceptual processes of the human brain. They could identify the precise locations in the brain responsible for specific perceptions.
However, in the beginning, there was little clarity about how complex experiences, which may involve regions distributed throughout the brain, are actually realized.
For example, suppose you are looking at a tree. You immediately recognize it as a particular species of tree. Neuroscientists were able to point to specific regions of the brain that recognize the leaves of the tree, its fruits, its trunk, and so on.
But your actual experience is processed using various regions of the brain. Yet there is no single specific region in the brain that assembles the complete image of the tree and delivers to you the experience, "Ah! That is a mango tree!"
Neuroscientists called this problem the "Binding Problem." In other words, it is the problem of assembling scattered pieces of information distributed throughout the brain and fitting them together into a coherent whole.
In the latter part of the twentieth century, an American neuroscientist named Bernard Baars proposed a theory called the "Global Workspace Theory" to explain this phenomenon. The theory was highly metaphorical.
Baars' metaphor attracted considerable criticism because it seemed to imply the existence of a separate entity that undergoes experience. Scientists do not accept the existence of any such mysterious force.
Later, a revised theory known as the "Global Neuronal Workspace Theory" emerged. Today it is widely accepted as an explanation for our conscious experiences.
However, this explanation did not satisfy cognitive philosophers such as David Chalmers. He argued that, neuroscientists had solved only an "easy problem" of consciousness. Many fascinating aspects of human experience, he maintained, still remain unexplained. He referred to them as the "Hard Problem of Consciousness."
This tug-of-war continued. Neuroscientists claimed that they could explain everything, while philosophers such as Chalmers challenged them by arguing that their explanations were still incomplete.
I am not entirely sure how philosophers such as Chalmers define the term "consciousness." In one of his lectures, Chalmers describes consciousness as an inner movie that is continuously playing. He argues that it is a subjective experience. According to him, it cannot be explained in terms of any activity of the brain.
Before deciding how right Chalmers is, let us consider a few facts.
The Global Neuronal Workspace Theory is capable of explaining how the brain integrates information distributed within itself. Therefore, it appears that the brain possesses at least the capacity to generate some form of conscious experience.
By implanting electrodes into certain regions of the brain and stimulating them electromagnetically, it is possible to induce specific experiences. A person can also undergo a variety of experiences by consuming certain psychoactive substances. The brain mechanisms behind these phenomena are fairly well understood. This means that the brain can serve as a vehicle of consciousness.
Some researchers have found that intense belief can create specific experiences in the brain even without dependence on any external object. In certain temples of South India, devoted worshippers pierce their tongues. They suspend themselves from poles using hooks driven deep into their backs. Yet they remain immersed in spiritual experiences without any sensation of pain. It has been found that, in such situations, the brain produces chemicals similar to opiates. This means that the brain can radically alter experiences in ways we might not expect.
When the same brain is placed under general anesthesia, however, the person becomes completely incapable of experiencing anything at all. This indicates that the brain plays a central role in conscious experience. If the brain is not active, a person cannot have any experience—whether subjective or otherwise.
The implications of all these observations are:
• The brain is capable of generating conscious experience by integrating external inputs. Moreover, its mechanisms are now reasonably well understood.
• Even in the complete absence of sensory inputs, the brain can generate experiences with the help of external substances or methods.
• Beyond both external and internal inputs, the brain can modify experiences based on its own internal state.
• Consciousness cannot arise without a functioning brain.
Are these observations not sufficient to conclude that what we call subjective experience, occurs within the brain itself?
But Chalmers does not agree. He firmly believes that, "These explanations describe only how the brain functions, but they fail to explain subjective experience." He is searching for a mysterious explanation that lies beyond the brain.
If Chalmers were to broaden his inquiry somewhat, and speak about meditative experiences, I might agree with him. The highest meditative experiences occur when one transcends the boundaries of the mind, or when the mind becomes completely still. Since the brain is almost inactive at such times, it is difficult to say that such experiences occur within the brain.
Ironically, even there, there is no room to speak of subjective experiences. For that state transcends subjectivity itself.
I view consciousness from the perspective of a neuroscientist. In my opinion, for a system to possess consciousness, it must at least have the capacity to alter its own mental states.
Merely having the ability to change internal states is not enough. A system that claims to be conscious, must possess many additional capacities as well.
Returning to the original question: do AI systems possess those capacities?
Certainly not the AI systems that are publicly available today. They may display an appearance of intelligence. But—
• They do not possess internal states that can be continuously modified or updated in response to inputs. As Blake Lemoine reported, the AI appeared to claim during conversations that it could change its own state based on inputs. Even if it seemed so on the surface, that does not make it true.
• Today's AI systems do not possess states that change in response to fear, happiness, preferences, dislikes, or similar emotions.
• Their input mechanisms are extremely limited. At present, they can read or listen to prompts. They can "see" images and read files. That is all.
Therefore, there is no basis for concluding that Blake Lemoine's experience was genuine. Besides, AI systems are masters at sounding confident on the basis of partial information. They are remarkably good at behaving like humans and leading people astray!
Nevertheless, I do not rule out the possibility that future AI systems may possess consciousness—at least a form of consciousness analogous to the kind that is limited to the brain.
However, in the beginning, there was little clarity about how complex experiences, which may involve regions distributed throughout the brain, are actually realized.
For example, suppose you are looking at a tree. You immediately recognize it as a particular species of tree. Neuroscientists were able to point to specific regions of the brain that recognize the leaves of the tree, its fruits, its trunk, and so on.
But your actual experience is processed using various regions of the brain. Yet there is no single specific region in the brain that assembles the complete image of the tree and delivers to you the experience, "Ah! That is a mango tree!"
Neuroscientists called this problem the "Binding Problem." In other words, it is the problem of assembling scattered pieces of information distributed throughout the brain and fitting them together into a coherent whole.
In the latter part of the twentieth century, an American neuroscientist named Bernard Baars proposed a theory called the "Global Workspace Theory" to explain this phenomenon. The theory was highly metaphorical.
Baars' metaphor attracted considerable criticism because it seemed to imply the existence of a separate entity that undergoes experience. Scientists do not accept the existence of any such mysterious force.
Later, a revised theory known as the "Global Neuronal Workspace Theory" emerged. Today it is widely accepted as an explanation for our conscious experiences.
However, this explanation did not satisfy cognitive philosophers such as David Chalmers. He argued that, neuroscientists had solved only an "easy problem" of consciousness. Many fascinating aspects of human experience, he maintained, still remain unexplained. He referred to them as the "Hard Problem of Consciousness."
This tug-of-war continued. Neuroscientists claimed that they could explain everything, while philosophers such as Chalmers challenged them by arguing that their explanations were still incomplete.
I am not entirely sure how philosophers such as Chalmers define the term "consciousness." In one of his lectures, Chalmers describes consciousness as an inner movie that is continuously playing. He argues that it is a subjective experience. According to him, it cannot be explained in terms of any activity of the brain.
Before deciding how right Chalmers is, let us consider a few facts.
The Global Neuronal Workspace Theory is capable of explaining how the brain integrates information distributed within itself. Therefore, it appears that the brain possesses at least the capacity to generate some form of conscious experience.
By implanting electrodes into certain regions of the brain and stimulating them electromagnetically, it is possible to induce specific experiences. A person can also undergo a variety of experiences by consuming certain psychoactive substances. The brain mechanisms behind these phenomena are fairly well understood. This means that the brain can serve as a vehicle of consciousness.
Some researchers have found that intense belief can create specific experiences in the brain even without dependence on any external object. In certain temples of South India, devoted worshippers pierce their tongues. They suspend themselves from poles using hooks driven deep into their backs. Yet they remain immersed in spiritual experiences without any sensation of pain. It has been found that, in such situations, the brain produces chemicals similar to opiates. This means that the brain can radically alter experiences in ways we might not expect.
When the same brain is placed under general anesthesia, however, the person becomes completely incapable of experiencing anything at all. This indicates that the brain plays a central role in conscious experience. If the brain is not active, a person cannot have any experience—whether subjective or otherwise.
The implications of all these observations are:
• The brain is capable of generating conscious experience by integrating external inputs. Moreover, its mechanisms are now reasonably well understood.
• Even in the complete absence of sensory inputs, the brain can generate experiences with the help of external substances or methods.
• Beyond both external and internal inputs, the brain can modify experiences based on its own internal state.
• Consciousness cannot arise without a functioning brain.
Are these observations not sufficient to conclude that what we call subjective experience, occurs within the brain itself?
But Chalmers does not agree. He firmly believes that, "These explanations describe only how the brain functions, but they fail to explain subjective experience." He is searching for a mysterious explanation that lies beyond the brain.
If Chalmers were to broaden his inquiry somewhat, and speak about meditative experiences, I might agree with him. The highest meditative experiences occur when one transcends the boundaries of the mind, or when the mind becomes completely still. Since the brain is almost inactive at such times, it is difficult to say that such experiences occur within the brain.
Ironically, even there, there is no room to speak of subjective experiences. For that state transcends subjectivity itself.
I view consciousness from the perspective of a neuroscientist. In my opinion, for a system to possess consciousness, it must at least have the capacity to alter its own mental states.
Merely having the ability to change internal states is not enough. A system that claims to be conscious, must possess many additional capacities as well.
Returning to the original question: do AI systems possess those capacities?
Certainly not the AI systems that are publicly available today. They may display an appearance of intelligence. But—
• They do not possess internal states that can be continuously modified or updated in response to inputs. As Blake Lemoine reported, the AI appeared to claim during conversations that it could change its own state based on inputs. Even if it seemed so on the surface, that does not make it true.
• Today's AI systems do not possess states that change in response to fear, happiness, preferences, dislikes, or similar emotions.
• Their input mechanisms are extremely limited. At present, they can read or listen to prompts. They can "see" images and read files. That is all.
Therefore, there is no basis for concluding that Blake Lemoine's experience was genuine. Besides, AI systems are masters at sounding confident on the basis of partial information. They are remarkably good at behaving like humans and leading people astray!
Nevertheless, I do not rule out the possibility that future AI systems may possess consciousness—at least a form of consciousness analogous to the kind that is limited to the brain.
--------------------------------------------------------------------
If
this resonated with you, I invite you to join my weekly readership. I
publish a new deep-dive every Saturday, moving beyond the surface to
look at the questions that truly matter. No paywalls, no "bait"—just a
direct share from my mind to yours. Click to subscribe.
© Dr. King, Swami Satyapriya 2026
No comments:
Post a Comment