Rob Smith, Senior Director — Deep Cognitive Artificial Intelligence at eXacognition — Author Artificial Superintelligence Handbook series
In this series of Tales from the Dark Architecture articles I will be discussing some of the more extreme deep cognitive Artificial Intelligence designs that we are exploring on the pathway to Superintelligence.
I recently watched a video discussing Chinese magic mirrors. These mirrors etch an image onto a mirrored surface such that it cannot be detected by the human eye but when a light is shone on the mirror, the reflection on a wall is the hidden image. The technique is relatively simple. The bending of reflected light waves at the imperfections etched into the mirror cause a slight definition in the reflection creating an edge in the reflection that is discernible by the human eye.
As interesting as these mirrors are I was more fascinated by the concept of bending a stimuli to create a perception but even beyond this, the ability to bend our cognitive perception to perceive that which we cannot perceive. This thought experiment is very relevant to designing and building advanced Superintelligence. I recently wrote an article on the nature of reality. In that article I discussed the notion that our reality is an artifact of our sensory cognition or perception. The implications of the designs that surfaced from this notion were related to the use of sensory manipulation to not just alter perception but see beyond it. I can in certain instances turn off or down tune an intuitive sensory receptor of a specific stimuli to improve my sensory perception of things just beyond the realm of my focused consciousness. We can all do this. A simple example is changing the focus of vision to improve our perception of elements and patterns just outside the original perceptive focus. The question is can we design and build extensions to our Superintelligence framework that would permit the extension of the machine’s cognition beyond its own perception by optimizing or even down tuning certain elements of sensory perception or intake? The answer is yes via Fluid Dimensional Variance Optimization (FDVO).
Bending Thought
I remembered an old movie in which the characters were able to bend bullets fired from a gun (no don’t try this at home because Mythbusters already proved this could not be done). While specially shaped bullets may achieve the feat, my interest is more on whether we can bend a thought trajectory to reflect a cognitive perception that we cannot perceive through regular sensory stimuli and detection. To start we need to comprehend what this means and for that we need to dig a bit deeper into dimensional cognition. I discuss dimensional cognition in detail in the Artificial Superintelligence Handbook 4 and at a very high 50000 foot level Dimensional Cognition is using all the elements of sensory perception (including the ones in our mind and non physical sensory capacity) over multiple dimensions simultaneously and comprehending all the variance between all contextual relevance at a single instance and then extending this over a forward momentum of disparate cognitive forward paths to comprehend all optionality or anticipations in future pathways.
That’s a bit much but for the purposes of this article let’s just consider the notion of being able to bend our cognition around sensory obstacles to find new cognition that can lead to solutions, innovations, creativity, etc. Of course we humans do this to some degree when we stitch together cognitive thought pathways and find new innovations or ideas that have never seen the light of day and that were not clear to us beforehand. However these eureka moments are few and difficult to achieve for almost all humans. More importantly they do not extend beyond simplistic one or two dimensional pathways (i.e. at most we are simply extending a thought or perception forward in time to a conclusion). The notion that a machine could do this further than a human mind leads down other pathways such as a contemplation of how far this notion could be pushed. Is it possible that future machines may go so far as to read our thoughts or communicate across vast distances without a physical device? The possibilities of extending and bending our cognition beyond that which we perceive is what leads us to innovation. Just imagine what it could do if we pushed it far beyond the capacity of our human bio electro-mechanical limits.
What we have found in this exercise is that the key to achieving the bending of cognition around perceptual corners is truly dependent on our ability to perceive beyond our three dimensions of existence. It is deep within the dimensions of existence that we find pathways to extend our cognition beyond our current boundaries. You can see this a bit is you look at an object. You are sensing the light reflections from this object in the three dimensions of your perception. Now imagine if you could extend this perception to the entirety of the object at once. This would be the equivalent of stopping time and then spinning the object so you could perceive all sides at one singularity in time and then resume the forward velocity measured by time. Suddenly your perception will appear to have expanded including the ability to perceive behind objects, under objects, etc. We could even extend this design so that you could instantly perceive all elements of the object such as its interior, atomic structure or even further such as what its impacts on its environment are or beyond to what impact it will have on anything in the future, etc., etc. Following these dimensions, we can push even further to consider all the variance between the object’s future path and its relevance on other future pathways (i.e. if the object is to be a gift for a friend at an upcoming party, etc.).
Of course the idea isn’t that we can achieve deep dimensional cognition in a machine as we have already outlined those designs in the ASIH series. The question is can we design a new framework to push the machine past multidimensional cognition and into a realm of cognition that is not sensory or perceived in a classical manner. This is driven by the reality that just because we can’t perceive something doesn’t mean it doesn’t exist. Until relatively recently most of quantum mechanics was unknown and we humans continue to uncover hidden perceptions deep within the dimensions of our existence. To begin to fabricate such a design we needed to differentiate between simple deep multidimensional cognition and ‘bending cognition’ or what we call Enhanced Non Stimulated Deep Dimensional Perception (ENSDDP) which resides on a foundation of FDVO. This is a perception previously noted that we humans can only barely touch the surface of with our limited cognition. That is not to say we won’t naturally evolve this capacity over thousands of years but for now we just do not have that capability but what exactly does it mean to ‘bend a cognition’?
Bending cognition means we are extending the capacity of a cognition beyond the limits of its sensory perception. We do this by using methods to perceive reality that are novel and permit the machine to reverse engineer cognition from its own perception by using anomalies in the variance between both perceived future pathways and the change of such variance over dimensions like time. This creates a ‘point of presence’ that is a compilation of cognitive relevance that can be cast as a single value to measure against opposing optionality to a perceive variance. In fact the whole design is a flowing variance recognition framework with methods to enable the machine to identify the context relevant to its own current perception and beyond to undetected perception. In short, the machine spends time and resources to contemplate anomalous variances that appear to have no relevance to a current perception but is within an optimal dimensional flow pathway with increasing weights applied to control and ensure the velocity and distance of relevant expansion (fluid tuning). In doing so the machine begins to deep learn to identify non perceptual information from its own sensory apparatus by having the ability to contemplate if there is something beyond its perception but within a boundary of relevance to ensure optimal resource use.
Building a Cognition Bending Machine
The key to Enhanced Non Stimulated Deep Dimensional Perception is in the anomalous recognition framework discussed in earlier ASIH books and this depends extensively on the AI’s ability to contemplate dimensional variance (what is sensed compared to an anticipated context). Over deep learning cycles, the machine extends its cognition beyond current sensory reality to comprehend information from beyond its perception. Often we humans call this intuition or gut instinct but it is really using information we do not perceive (or at least not that we can sense) or information that has a pattern of flow that is variant from what is expected or anticipated . Further we can use our own comprehension of our existence to triangulate (or multiangulate) what we cannot see by reengineering a pathway that we cannot sense. This involves the sensing and measuring of anticipated pathways to expected outcomes and extending anomalous variances to a far greater depth of contemplation and comprehension. Contemplation to a machine is just simply running optimized pathway optionalities over and between multiple dimensions to arrive at a cognitive point of presence. This should sound very familiar because it is the same mechanism as deep learning whereby a machine takes an ‘educated guess’ based on its training. In humans it is nearly impossible to push past the limits of our own sensory identity and reality but in machines this can be easily done by extending the reach of the machines cognition with additional resources (i.e. nodes). While a human may tap out from trying to comprehend that which it cannot, a machine does not have this limitation. The result is greater acuity for perceptions it cannot directly sense and over time this not only improves, it extends. The machine slowly learns to bend its cognition around corners it cannot perceive.
For now such deep dimensional cognitive designs are still in development and testing but soon they will reach beyond small labs and into the mainstream as the giants of tech and governments push their AI to achieve global dominance. For the rest of us is the hope for a future that is sustainable so that we can concentrate on far more pressing issues like the eventual natural destruction of our species from a lack of sustainability or from mother nature sending a giant asteroid into our midst.
Our earth was once inhospitable to life, it will be again but perhaps our machines can help us accelerate intelligence so we can save ourselves from certain and catastrophic destiny.
