There is a recent paper in Trends in Neurosciences, The feasibility of artificial consciousness through the lens of neuroscience, with the abstract, “Interactions with large language models (LLMs) have led to the suggestion that these models may soon be conscious. From the perspective of neuroscience, this position is difficult to defend. For one, the inputs to LLMs lack the embodied, embedded information content characteristic of our sensory contact with the world around us. Secondly, the architectures of present-day artificial intelligence algorithms are missing key features of the thalamocortical system that have been linked to conscious awareness in mammals. Finally, the evolutionary and developmental trajectories that led to the emergence of living conscious organisms arguably have no parallels in artificial systems as envisioned today. The existence of living organisms depends on their actions and their survival is intricately linked to multi-level cellular, inter-cellular, and organismal processes culminating in agency and consciousness.”
For one, the inputs to LLMs lack the embodied, embedded information content characteristic of our sensory contact with the world around us.
It is possible for an individual to be conscious, without the embodied, embedded information content of sensations. There is incoming sensory information. There is how the brain defines, or interprets the information. Interpretation is what becomes of the information. The color is the color because of what the brain says it is. This means that the brain already has configurations, where information comes into, to shape [or interpret] it for experiences. The configuration is often previously learned. This means that the familiarity with [or definition of] a smell is a result of what was acquired about it, then organized for subsequent use.
There is the ability to smell. There is the ability to know what the smell is. Someone may go to a new place and smell something and not be sure of what it is. What is often more important is to know what that smell is beyond the basic ability to smell. If smell, sight, touch or others, cannot be known, risks abound. Knowing includes ‘association’.
Interpretation may play out in thoughts, memories, emotions, without sensory contact with the world [responsible for those] in a moment.
Generative AI has this ability to know, though it does not have the ability to sense. This slice is not zero, for consciousness.
Secondly, the architectures of present-day artificial intelligence algorithms are missing key features of the thalamocortical system that have been linked to conscious awareness in mammals.
The thalamus integrates sensory information. The cerebral cortex does interpretation. How? The thalamus has several nucleus or clusters of neurons, with separate functions. But how do these nuclei work to do what they do?
Conceptually, the electrical and chemical impulses of nuclei are in loops. These loops have formations, or rations of chemical impulses, which is how information is organized. The loops or sets of impulses have several features. Their ultimate function is to make something of what comes in and distribute. It applies to the cerebral cortex.
The key elements are electrical and chemical impulses, whose sets organize information or structure them into experiences, including for the sense of self, attention and intentionality.
What these impulses do as sets is to know, playing their part and sharing. Generative AI does not have impulses, but its neural networks can know, giving it a non-zero measure compared to human consciousness.
Finally, the evolutionary and developmental trajectories that led to the emergence of living conscious organisms arguably have no parallels in artificial systems as envisioned today.
AI scooped information at an advanced stage of human civilization. It does not have to experience things from scratch. What AI has, is like, the manual of human totality. It may not have agency yet, but it is in possession of the manual. Humans have better quality of life, adapting advances that were not individually experienced, but deemed credible. If generative AI were to study humans and humans were to study AI, AI may figure out more about humans. If humans were to study animals and animals were to study humans. Humans would figure out more.
The existence of living organisms depends on their actions and their survival is intricately linked to multi-level cellular, inter-cellular, and organismal processes culminating in agency and consciousness.
The ability to know: know of the self, have intentionality, and experiences. Knowing is the factor, not just to exist. AI does not experience. But it can be ascribed to it that it has outputs that know about those experiences, giving it a foot in the door of human consciousness and intelligence.
—
This Post is republished on Medium.
—
Photo credit: iStock
