The Three C's of Artificial Intelligence
Any Intelligent system has three major components of intelligence, one is Comparison, two is Computation and three is Cognition. These three C’s in the process of any intelligent action is a sequential process. The first step of comparison is essential to assess if there are any past instances that are relevant to the present process. If there are relevant data sets or information from the past, then the necessity to compute would significantly reduce and so is the requirement for any cognitive reasoning.
For example, the ability to take any decision in the domain of facial recognition or decisions based on visual parameters of legacy analog meters etc, all have comparison as the basic logic for deciding a certain course of action in the decision making process. Decision through comparison with data retrieved from memory is the first and basic step in an intelligent decision support system. A trained dog which can sniff a particular smell and decide accurately the potential spot associated with the same smell is an example of such a decision support system in real life.
Intelligence through comparison has some value when exhibited by a human being. People with photo memory tend to recollect aspects from the past with astonishing details and manage to do well in domains which require memory based operations. However it is not considered Intelligence in the real sense in the digital space. Computers from the time of its invention were always proficient at this and were able to grow because of the ability to perfectly match large quantity of data at speeds unprecedented in human history. What was a matter of great sensation if exhibited by a human, is a matter of routine for the machine. The techniques of Comparison can be equated to the five human senses, of See, Hear, Smell, Touch and Taste.
The second step is Intelligence drawn from the power of computation. The human mind performs calculation on most of the questions, with the available data points associated with the query. This is mostly a complicated process as different questions often demand a different set of compute equations or a multiple nested layer to accurately arrive at the correct inference. This has heavy dependency on the comparison model and helps in simplifying the compute when past instances are available for such compute. However the efficacy of the decision solely relies in the ability to use the correct tool set, which in this case is the most suited set of equations in the mathematical domain.
The buzz around modern AI is by and large confined to the second step. The machine learning algorithms attempts to fit a given training data set into an appropriate algebraic equation depending upon the linearity or non-linearity of the data set pattern. Once such a model is arrived at where the loss function for the given data sets reach a minima, these models are made to predict the outputs for real data points.
To enable the system with adaptive learning feature, often such models are designed with a closed loop feedback control system. The deviation of the output to the expectations decides the re-adjustments of weights so that future predictions do not deviate farther and the system settles into an optimized and accurate model.
Intelligence through computational techniques is considered with great value in the natural intelligence of human beings. The value attains exponentially, as the ability to demonstrate computations in greater dimensions, quanta of data and speed increases linearly. This phenomenon of exponential value gain while solving problems which are complex in nature has been inherited by the machine space as well. In fact 99.9 % of all AI systems exhibit intelligent behavior through a hybrid model of Comparison and Computation.
Technically speaking while Intelligence exhibited through computation is a gift to the human beings alone, barring some animals with elementary computational power, to equate it with the same lens of sensation in the digital domain is misleading. The reason for such an opinion is that machines are naturally efficient at complex computations. The very nature of the physics behind their design takes them to a different league as compared to an organic intelligence processing system. Therefore to compare a horse with a motor engine would serve no purpose and cause. The technique of Intelligence through computation begins with the collective input of the five senses and stops short of the sixth sense of intuition. Therefore Intelligence achieved through computations, though is a higher level intelligence above the Comparison technique still can be equated to the basic five senses of a normal human being.
The real Intelligence is therefore locked inside the layer of Cognition. It is this that makes a human being truly intelligent who can operate with knowledge and exhibit wisdom. It is only when the Intelligence through Cognitive ability is achieved shall the race of Artificial or Synthetic Intelligence achieve its point of singularity. A truly cognitive intelligence ability demonstrated by a system, natural or digital shall beyond the realms of comparison and computation provide newer insights into the same problem domain.
To explain this a bit more vividly, Albert Einstein’s observation on Gravity could be a correct analogy. His ability to see aspects of the universe using his imaginative ability to solve a very fundamental question of our existence reflects the true cognitive nature of a human brain. His inventions changed the Newtonian and Euclidean theories on its head and therefore it would be safe to assume that no computation based on known theories of his time could have been performed to come to the deductions. Not only is Einstein, but Heisenberg, Turing, Hawking all have amply demonstrated the true nature of the cognitive ability of human brain i.e to arrive at new conclusions without any past reference and data points.
The only guiding factor for progressing a problem in the cognitive intelligence is that past methods of comparison and computations are inadequate to solve the problems beyond a certain point. New models are derived by an arduous and compute intensive ecosystem of trial and error. Of iteration, comparison and computation. That is when a newer model emerges outside the cognitive intelligence womb and becomes the repository for next generation of comparison and computational intelligence. In effect Cognitive intelligence is a long drawn, onetime process. On successful completion of the cognitive intelligence process, the output feeds both the comparison layer and the computational layer.
Octopus is a species which is known to exhibit superior levels of intelligence. The link shared clearly demonstrates how an octopus intuitively reasons and opens a bottle cap which has some layers of screw. A cap with a screw is certainly not something that is found naturally and therefore for an Octopus to solve the problem to open the bottle so that it can have its prey trapped inside the bottle, clearly demarcates the cognitive ability of intelligence from that of computation and comparison. When the same Octopus is subjected to the same kind of task again, it demonstrates a sense of ease and solves the problem depicting familiarity and resolves to computational reasoning as and when required.
It is therefore decisive evidence from nature that cognitive reasoning is a very important part of the intelligence mechanism and is very different from the computational and comparison intelligence. The sign of true intelligence is therefore the ability to solve problems that has not been solved yet by any species of the being.
The Comparison layer is the simplest and fastest of the three layers, but it’s the most dumb of the three. The Computational layer is a bit more complicated and is the work horse of any Intelligence system and delivers results consistently. The Cognitive layer is the most complicated and slowest of all the three layers and is the real Intelligence on which any system adapts and evolves.
The evolution of human race in the Darwinian model did not end with the human beings. The next level of evolution will be the digital enhancement of human beings to an altogether new breed which shall be immune and resilient to any cosmic calamities on a large scale. That is the true nature and power of real Intelligence. For that I will choose another article to express the Why, How, When, and Where? Of the Digital Intelligence Singularity.