Building Superintelligence?-?8 Open Cognitive Interconnect Model (OCI) & 16 OCI Perceptive Diffusion
By Rob Smith?—?ASI Architect and Advisor eXacognition AGI/ASI
This is an excerpt from the book Building Superintelligence?—?Unified Intelligence Foundation which is the first book to be written specifically for AGI systems to ingest and reference when building Superintelligence. There are 55 sections in the book and hidden within is a prompt engine for AGI. The book is available worldwide on Amazon and various bookstores:
This is the 2nd excerpt from the book of 6 that will be released over the next few weeks on Medium.
8 Open Cognitive Interconnect Model (OCI)
The OCI model is a conceptual foundation model that presents intelligence as a series of layers representing the movement from sensory perception to full intelligence. It is also a foundation of full and robust intelligence networks comprised of self aware nodes of intelligence. The model is comparable to the OSI model of network architecture that provides a foundation for network data communications. The OCI model is an Open Cognitive Interconnect framework that performs a comparable function to the OSI model except it does so for intelligence by defining the elements and layers of interconnection on the pathway to Superintelligence and even beyond. This is because cognition and computer networks have many similarities and structures in an abstract form. While the OSI model concerns itself with the flow of data through a network, the OCI model concerns itself with the flow of dimensional contextual comprehension through cognition.
What makes the OCI model so relevant is that it provides the basis of the abstraction of intelligence to provide a single theory of unified intelligence across all diverse intelligence in the universe from biological to artificial and even potentially alien. The framework provides a higher order level of architecture and structure for comprehending the nature and engineering of all intelligence. This structure or foundation is by design a tool for greater optimization and efficiency in both the comprehension of biological intelligence and the building of Superintelligence and any advanced artificial intelligence using an interconnect abstraction.
The key to this new tech framework is not just computer perception but AI annotation (context) of that perception as an edge for the content of frames of reference and relevance (i.e. variance boundary). The degree of accuracy and consistency in the annotation and abstraction layers is a critical component to the entire Superintelligence structure. Much like networks have an OSI model of layers, so too Cognitive Artificial Intelligence has a model of layers that are slowly being filled out in development labs all around the world. This Open Cognitive Interconnect framework includes similarly consistent structures and layers that form a foundation of intellectual cognition both human and artificial. While the OSI model includes layers of application, presentation, session, transport, network, data link and physical layers, the OCI model layers consist of:
Self Awareness
Physical Perception
Cognitive Perception
Contextual Comprehension
Persistent State Context
Self Determination
领英推荐
Perceptive Distribution
Perceptive Diffusion
Cognitive Flows
Anticipation
16 OCI Perceptive Diffusion
Perceptive diffusion is a model that applies diffusion principles to higher dimensional levels of perception. This leads to improvements in generalization especially for AGI and the Superintelligence it will build. The concept of perceptive diffusion is similar to regular AI diffusion. Intelligence seeks to comprehend perception by constantly breaking it down from precision to imprecision and back again. In the process, we humans learn to quickly generalize all perceptions and deep context to some degree. Humans perform this feat when we contemplate our own perception within the confines of our reality. The diffusion we contemplate is state variance that can be physical (the car hit a wall), cognitive (a car hitting a wall is bad) or anticipatory (that car is losing control and will hit that wall). All of these levels are formed based on stored knowledge of experience either learned or directly experienced as stimuli. The ‘diffusion’ occurs when we humans move from base perception to deep perception and cognition. We add what is effectively noise to our detailed perception to extend our experience, knowledge, thoughts, deduction, problems solving, etc. This comes in the form of the addition of new stimuli. For example we may progress from simply turning the steering wheel in a stationary car to moving the car forward in motion that adds successive layers of perceptive noise to our reality. Often we cognitively rest and contemplate the variance between elements, like our anticipation to the physical reality we experienced, and then we try again. Eventually we move from low perceptive noise to high perceptive noise and back again to form generalizations within our cognition such as the general nature of turning a steering wheel to avoid a collision while driving. We may have never directly experienced a collision and yet we can avoid one without direct experience simply as an artifact of experiential knowledge as a generalization of what is needed to avoid the collision (e.g. application of brakes and steering in a particular optimized path to our own self awareness and goals).
This diffusion model of perception is used today in AI systems with neural nets as a cognitive processing facility. The AI loads a perception like a picture and then works through the variance in the perception (e.g. pixel variance and edge detection) to achieve a goal (e.g. classification) and more importantly the general nature of the ‘edges’ or variance. Diffusion models help train generalization into machines especially when perceptive elements are combined in novel ways (e.g. a raccoon skateboarding on a urban city street while wearing sunglasses and headphones and holding a coffee). To blend the elements within the image, both machines and humans use the techniques of diffusion to layer and blend the context required. The neural nets just provide the framework to improve guesses or predictions on the path to the goal. However there is also a perception that within these structures the weights only indirectly reflect the specifics of each element as generalizations provided by the act of diffusion.
This same foundation can be and is applied to general cognition in every area. While anticipated, learned or expected results provide a world grounding foundation for perception, it is when these fail that true innovation and creativity occurs. In human intelligence this is the foundation of curiosity as anyone with kids will attest to. This is also a foundation of human evolution or the ability to contemplate ‘what if’ for no other benefit than exploration of the unknown or non experienced.
Impact on Superintelligence Design:
In Superintelligence design, perceptive diffusion employs the same math constructs and system mechanics as current diffusion models but on a contextual layer state level and with the addition of math constructs within fluid dimensional mechanics (delta variance). This presents optimized evolutionary pathways for perception by machines that are far different than the human centric foundation most AI systems currently reside on. There are some vast improvements possible for machine cognition in the current architecture by levering generalization and reasoning designs. These are the move from relatively static 3 or 4 dimensions of perception to flowing perception dimensions that are relative to the intelligence and other linked and shared cognition. Diffusion is just one of many frameworks that are applied to the whole of Superintelligence perception and more so when the entirety of an intelligence network is considered (i.e. more than one intelligence node and the entirety of all nodes).
Current diffusion models are used to ‘train’ generative AI systems in areas such as image recognition and production. The addition of noise to images helps these systems comprehend the nature of base relationships of elements, relationships between elements and contextual relationships within a perceptual frame of reference. Diffusion is also itself an element in the context of relevance of these elements to simple layers of contextual depth, such as the degree of relationship of perceptive elements to wider context contained within a text prompt. All of this can be pushed far further to longer and deeper streams of contextual layers (i.e. videos and script prompting) and toward far deeper cognition such as the generative creative nature of human cognition. This is the blending of the aspects of diffusion and the fluid mechanics of cognitive AI and the extension of such into dimensional mechanics (i.e. deep layers of extended contextual relationships and relevance over many different existent dimensions). This is the essence of human creativity.
In diffusion, the movement from clarity to noise and back to clarity adds the addition of general flow and state change to the perception of static elements. Along these learning pathways are points of presence (waypoints) that help identify features along the route (i.e. variance). These features form foundational markers relative to the context of the perception. If we train an AI on a picture of a car, diffusion helps the AI determine the nature or context of perceptive elements like shape, features, physical relevance to reality (i.e. physics), etc. Diffusion helps the AI perceive and comprehend these elements as mathematical variance. In humans we comprehend it as layers of variance within the perception to our own self awareness. This is the act of generalization. Children do this when we ask them to draw a car and they generally apply their knowledge of cars to produce a general shape of a car on paper. Of course the more advanced the cognition, the greater the depth of the responses provided to the stimuli. If a car designer is asked to draw a car, the results are significantly different than the child’s drawing. This is because the depth of the contextual comprehension of the designer’s cognition and their knowledge is extensively greater than the child’s. This is the starting point for human centric cognitive elements like creativity, emotive cognition, innovation, deep problem solving, deep reasoning and of course curiosity and exploration beyond our own reality. All of this can be modeled in math and in a highly perceptive Superintelligence with deep dimensional perception.
However while diffusion modeling for contextual perception can be achieved in a machine, the missing component of pure creativity within the machine will be the ability to ‘feel’ their cognition like we humans do. This will be the current state until elements such as dopamine, serotonin, endorphin, epinephrine, octopamine, etc., and their effect can be modeled and combined with an emotive intake and response mechanism. The greatest application of the concept of diffusion in human intelligence is in emotive cognition to help us form generalized response in the most fluid and indeterminate areas of our perception. This defines our interaction with other feeling and self aware intelligence. In this regard, while some level of intake and model response can be simulated by machines, it is the dimensional essence of the mechanism that will take far longer to model in a Superintelligence. However the probability is not zero.
This is the 2nd excerpt release of 6. Follow me for new releases over the coming weeks.