Why AGI and cognitive science is a wicked hard problem
OK, let's get serious for a nanosecond at least
Artificial intelligence and cognitive science will make major advances when they understand that the human mind is composed of a hierarchy of information systems, from neurons all the way down to small molecule signaling and including our large sensory subsystems.
Our body is the embodiment of our mind.
In my quest to unite the physics world of thermodynamics with the world of information, I was led to the study of neurobiology and then to the study of molecular information systems in general.
This led to a realization that every cell in our body(the superorganism) is a dependent organism, each running the same information/thermodynamic base model as our complete body. Then, within and among each of the cells operate multitudes of additional information systems. Some are dependent, enclosed in their cellular superorganism, and some of them, like the immune system, act as distributed agents within our larger superorganism body.
These can be classified roughly as:
– Neuronal - The highway system
– Proteomic - The infrastructure agents
– Hormonal - The cellular democracy system
– The DNA subsystem - And all it’s enabling and regulating subsystems
– Small molecule signaling - Ion channels, kinases etc.
Each of these classes of information systems operates within and among every cell in our body, including our brain. For example, the mitochondria that produce ATP and provide the energy to run our complex large biology systems were in fact independent-living bacteria that took up residence as dependent organisms within our cells, but still retain their own inheritable DNA subsystems.
As another example, our hormonal information systems output hormones, which express the needs and desires of each of our cells. Hormones circulate throughout our body, acting as votes in a kind of body-wide democracy. That hormones have a profound effect on the operation of our brain is beyond dispute. Therefore, any model of brain function must also consider the operation of the hormonal system that is influencing the operation of the neurons.
In addition, our microbiomes, consisting of bacteria, viruses, and fungi, count far more cells and vastly more DNA than our actual human biome. Are we just the vehicles that these lifeforms use as a way of getting around town? When our out-of-control gut candida gets hungry, it sends hormones to the brain that, in the presence of an advertising information virus, triggers us to suddenly have an irresistible urge for pizza. Like trained dogs, we get up, put on our clothes, find our money, get our keys, get in the car drive to the pizza parlor. They live in us because humans are much easier to train than dogs.
Perhaps the complexity of trillions of interconnected, multi-modal information systems is why mechanistic cognitive scientists have such a hard time reliably explaining behavior based on a few electrodes in the gray matter. It is also an indicator that true GAI is impossible based on computational thermodynamics considerations.
We need a new way of looking at the relationship of AI and human minds. Instead of imitating, AI needs to learn how to communicate better.
The exploration and explication of all this is the subject of future articles. Next up is the most important presentation of the foundational model of thermodynamic/information lifeforms, as I will describe in my tongue-twisting Theory of Constructal Infonomics in a separate Substack.
Thanks for reading I Think I See! Subscribe for free to receive new posts and support my work.