—By David Stephen
There is a recent comment by a professor, “The brain of a house cat has about 800 million neurons. You have to multiply that by 2,000 to get to the number of synapses, or the connections between neurons, which is the equivalent of the number of parameters in an LLM. So maybe we are at the size of a cat. But why aren’t those systems as smart as a cat? A cat can remember, can understand the physical world, can plan complex actions, can do some level of reasoning—actually much better than the biggest LLMs. That tells you we are missing something conceptually big to get machines to be as intelligent as animals and humans.”
This comparison of the brain to LLMs, like the earth to mars, simply because mars does not support life is off the mark. The brain is known to be unique among all organs, because it bears key landmarks of the mind. The mind is theorized to be responsible for all information functions of the brain, including memory, modulation, emotion, perception, sensation, intelligence, creativity, thought, feeling, planning, integration, interpretation, observation, curiosity and so forth.
Synapses are theorized to be special because they are the closest architecture—of the brain or body—for the mind. Conceptually, the mind is the collection of all the electrical and chemical impulses of neurons, with their features and interactions, in sets. Everything else in the cranium is the brain. Outside the cranium, everything else is the body. Impulses, with their features and interactions are the mind, providing a mind-body distinction.
Neurons and their synapses—are part of the brain that directly shape the mind, but they—are not parts of the mind. The mind is available outside of the brain, with electrical and chemical impulses of the peripheral nervous system. However, most determinations for functions are made in mind configurations within the cranium.
There are several things the mind does that are obtainable elsewhere. There are organisms without a brain that can do some of the things the mind does. There are organisms with smaller brains than humans that can do the same. Because of the varied functions of mind, diverse comparisons cannot be ruled out.
There is no function of the mind that is not predicated on memory. Regulation of internal senses, for example, kidneys or liver, are done within limits and extents of information—or memory, and to create alerts for anomalies. Simply, it is the memory of normal that the mind uses to gauge inputs from internal senses then overdrive when amiss.
Thoughts are done with what is in memory, so are feelings—or information on what is available, as states. Since a principal function of the mind is memory, wherever memory can be found can be compared to the mind, regardless of differences.
Artificial memory is available on paper, walls, sculptures, footprints and so forth. The materials are unable to do anything with them, but organisms use those memories to make inferences. Digital too had memory, but could do nothing with it, until AI came by.
Digital, with video and audio, possesses a memory that is sharper than any organism. But this memory could not be used to generate new memories within, or do much, until LLMs.
This means that memory can be a function, but it is what acts on it, that counts. The option that LLMs have to act on is digital, so the limits of LLMs are the limits of digital. Comparing a house cat with its sensory sphere in the physical to LLMs, limited by digital, is a kind of error.
The human mind is more advanced than the animal mind because of extended functions and extended qualifiers. Simply, there are more functions in the human mind and there are more ranges of qualifiers that act on those functions. Language, for example, is an extra function, acted on by more qualifiers. The lack of qualifiers for information in materials like paper, walls and others, is a refutation of panpsychism.
LLMs are able to qualify aspects of digital memory, leading to some of the accuracy they show in outputs. Some people can dismiss large language models as statistics, but those statistics are resulting in some of the same things that the human mind does—on digital.
Qualifiers of functions include attention, awareness, free will or intent, self or subjectivity, distribution and so forth. Consciousness, as a super qualifier, is a collection of all qualifiers. While the qualifiers in the sentient mind are far extensive, LLMs as qualifiers, to digital memory, have tad parallels.
Attention is obtainable when with a prompt and result. Awareness, like peripheral vision, is obtainable where LLMs acknowledge information from a prior prompt, while answering the current one. While they have no subjective experience, LLMs have a rudimentary sense of being. Though programmed, they answer in the first object, like in the first person. There is also a form of intent or free will by LLMs, where they are able to answer as told, in a voice, with a video setup, in a writing style or image type. Though prompted, they exhibit a weak form of intent or control in their outcomes.
As they get better, arguments that they are nothing may not count, if they cannot be compared with the qualifiers they are deploying to digital memory. Digital already percolated human productivity and social activity. AI can qualify what is within digital, with some nefarious results like deepfake voices, videos, images, misinformation and so forth. Dismissing LLMs intelligence qualification, in a digital-driven world, for questions of AI safety, could be another kind of error.
