Project A.G.INT. is combining knowledge across the fields of computer science, psychology, and neuroscience, in an effort to discover the technical requirements for consciousness, using the brains of animals (mostly humans) as reference.
1) Create an AGI that is on par with humans.
2) Create a general model/definition of consciousness.
3) Create AGI that is suitable for practical use.
Project A.G.INT. combines 2 approaches to AGI. The first uses psychological models and observations of the mind as a basis for what the model needs to be capable of, then takes those requirements and figures out how to implement them on the neural level.
With natural minds being the only available examples of cognitive models, Project A.G.INT. limits designs to only what can exist in the real world. This means stuff like limiting fully connected parts in the network, since for the most part neurons can only send data to their neighbors.
The second approach looks at the fundamentals of data processing. Things like what it means for information to be observed and perceived, data representation in a network, and maintaining a single entity over a network. These fundamentals are then used to set design rules and boundaries for the cognitive model, as every part of the model must uphold and abide by these fundamentals.
Lastly, for guidance, Project A.G.INT. looks at the nervous system of animals. Current knowledge and research of the human brain is used to determine where certain functionality should exist within the cognitive model, as well as the types of neurons and other processing mechanisms that can be used in the model. The nervous system of simple organisms, and the history of the nervous system and its evolution are used to see where to start building the cognitive model from, and how its different parts have been added over time.
Using animal brains as reference, Project A.G.INT. looks at the form that data exists in and the processing that is done at every point in the network, from sensory input to nervous output (e.g. muscles).
Project A.G.INT.'s first goal is to create a copy of the human brain's cognitive model. Until consciousness is better understood, Project A.G.INT. is trying to stay as close to the available example as possible.
The second goal is to create a general model/definition of consciousness, one that is as simple as possible, which will mean 2 things:
And
This second part means that systems that have been unconsidered for cognition may be able to be recognized as conscious.
The third goal is to use the knowledge of cognition to create simple AGI that can be used for task specific purposes, similar to artificial narrow intelligence.
Project A.G.INT. was started with the motivation to create an extremely intelligent being, whom could drive and accelerate advancements in science and technology.
Project A.G.INT. is not a machine learning / deep learning model. While being an important part, and having a role in Project A.G.INT.'s cognitive model, learning and using learned knowledge does not cover the full scope of consciousness. Other things, such as the structurally hardwired way that data (like that from the senses) is processed, also have large roles. So while the model of Project A.G.INT. technically falls under the classification of machine learning, it should not be thought of as such.
