Trustworthy Artificial Intelligence

Trustworthy artificial intelligence is an incessantly debated and strenuously researched hot topic. It stems from the DOUBT on the intention of machines that would develop intelligence. Human beings are trying to spawn non-living beings that have much higher potential than human beings.

A machine is devoid of life - a quality that distinguishes matter having biological processes from other forms of matter. Machines have a faster, tireless, and more efficient organisation of systems for functioning. With the right nurturing and sustenance, they can be immortal - exist without death or decay.

Immortality is desired by each form of life and reproduction is a result of life's longing for itself.

Human beings strive to empower a machine with the quintessential capability of human beings - intelligence. But, how did intelligence develop in the first place? Perhaps this question requires an answer to find a solution to the conundrum of developing trustworthy AI.

The approach entailed in this article is philosophical and self-exploratory. All assertions are mere propositions that need to be self-verified and validated against the touchstone of one's own understanding and knowledge through critical thinking, reasoning, and analysis.

What distinguishes human beings from other forms of living beings? The widely accepted theory of evolution is yet to stand the test of time to be deemed as a law. But evolution proposes that a cognitive revolution in our ape-like distant cousins triggered an ability in our brain to "imagine". A few physicochemical changes in the human brain led to the faculty of imagination. It is proposed that this unique potential distinguishes an animal consciousness from a human consciousness. Human beings have conceived imaginative realities such as language, currency, religion, cities, state, and political ideologies that reside in the collective imagination of the people. Human beings believe in such collective imagination to discover a brotherhood or camaraderie for mutual cooperation, perhaps unlike any other species in the animal kingdom. This has helped human beings break the threshold of social groups and ensure large-scale cooperation for the current progress of the species.

Human beings over the past 70,000 years have evolved so fast and climbed to the top of the food chain without allowing time for other beings to adapt to find a balance or harmony in existence. The cruelty and dangerous nature inherent in human beings perhaps must have resulted from the fear of being hunted and preyed upon when human predecessors were in the middle of the food chain. Behavior that is based on an assumption about the intention of the other rather than understanding and knowing - this is perhaps the source of the desire for dominance as well.

When human beings try to design a new form of being mirroring their own self, perhaps they must know their self. This is felt to be imperative if there must be sustainability in nature when new beings flourish and dominate the order in nature.

The term "self" is used intentionally with a profound meaning - humans have designed a body for machines that is significantly different from our own biological bodies. But, the faculty of imagining, understanding, and realising, which they strive to inculcate in the machines are pretty much mirrored from the self.

Why are human beings having an unsustainable relationship with the other three orders of nature - the physical order, the biological order, and the animal order? Without properly addressing this, we will not be able to design harmonious artificial intelligence. Such artificial intelligence will precipitate fundamental flaws in human beings such as lack of trustworthiness, lack of respect, malevolence to cause harm to others, complexity that renders its behavior inexplicable, injustice or lack of equity/fairness, disharmony, and unsustainability. It must be clearly understood that the arbitrariness in the behavior of AI is a result of arbitrariness in human beings.

And we will suspect the machines to do to us, what we did to the other forms of existence when we rose to the top of the food chain.

It is proposed that human beings should realise the right understanding of the "self" and its potential that they are going to mirror in the machines.

It is felt that current learning in LLMs is purely based on the context of behavior rather than having the right understanding of the concept that has to be learnt. For instance, an NLP is context-aware when it learns the possible meaning of a prompt. This is possible when the model is trained on large corpora of information from which it assimilates the knowledge.

Language is supposed to be a means for not just unambiguous communication. For its effectiveness, it must be precise as well. The loss of information must be tending to absolute zero. Linguists might agree when it is proposed that the meaning of a word must be absolute. For instance, the meaning of "respect" needs to be unconditional and universal. The purpose and relevance of value such as respect is lost when it becomes conditional to the behavioral context. Respect is the right evaluation of another human being. The purpose of respect is to understand and mark the other as an equal with varying competence and to take up the responsibility to build the competence for mutual complementariness. When a person learns the meaning of respect from a prose in which the respect is embedded with a preceding statement that indicates a behavior of a person being respected, the wrong understanding that respect must be based on behavior is learnt. Gradually, this leads to the wrong evaluation and assumption that the purpose of evaluation of the other is to judge them based on their competence. This aspect of LLMs must be revisited perhaps at a later point as this would digress from the main thread of discussion at hand.

The current predicament of human beings is caused by behavior based on a lack of right understanding. When a human being is born, we consider it to be a co-existence of a body and a self. While the amount of understanding at the point of birth is unclear, it is assumed that the self at the point of birth has formed no concepts and has absolutely no understanding. The body may have involuntary abilities as in any perfectly harmonised organisation of interconnected systems. What is genetically inherited is also uncertain. However, it is proposed that it possesses no residual knowledge, and no communicative ability. Even the muscular reflexes are not based on understanding.

With a clear understanding of the distinction between the self and the body of a human being, it can be seen that it is the self that develops abilities based on its potential for learning and imagination.

The role of the body, amongst other systemic functions, is that of an instrument for the self to sense and respond to its immediate surroundings. The perceived information is passed on to the self. The self recognises the information and fulfills any need for sustenance of the body. Other activities of the self also include knowing, assuming, understanding, and thinking.

It is proposed that the human self is in a continuous state of imagination. The self nurtures desires and continuously thinks about means to fulfill the desire. The thoughts are spurred by some expectations. Desire is the activity of imaging, where one makes an image of something in the self. Thought is the ability to compare and analyse various possibilities to fulfil a desire. Expectation is the ability to select based on taste. Based on the taste of the self, they select a possible outcome, which they assume would manifest in their selves when a desire is fulfilled. Imagination of a self is expressed to the outside world in the form of behavior with fellow beings and is reflected in nature through their work.

Consciousness is the sum total of the realisations, understanding, knowledge, desires, thoughts, and expectations that develop in the self over time since birth. It is proposed that while the potential of self in every human is universally the same, the body differs in capabilities. The consciousness distinguishes each human being from the other.

It is proposed that there are three possible sources of motivation for imagination:

1. Preconditioning - When the beliefs, notions, norms, ideas, views, dictums, or goals prevailing in the family, or in the society which have been assumed without knowing, influence our imagination, it is termed preconditioning.

2. Sensation - When the sensory perception from the five sense organs becomes the source of imagination, it is termed sensation. What we learn from others or from an environment through sensory perception accounts to sensation.

3. Natural Acceptance - When the feelings of innate humanity are considered as the source of motivation, it is termed natural acceptance. It is proposed that such imagination should be acceptable to any human being universally irrespective of criteria such as time, space, gender, or caste.

When the source of imagination is predominantly based on preconditioning or sensation, the happiness of the self is enslaved to external sources. Some preconditioning or sensation may also be right at times. However, if not self-verified on the basis of natural acceptance and validated by practicing it living to ensure mutual happiness, it is just like enslavement. Similarly, while sensation in itself is essential for nurturing and sustenance of the body, associating happiness with the feeling of favourable sensation leads to enslavement.

The imagination must be motivated by self-verification based on natural acceptance - asking the question of whether the imagination is naturally acceptable to any human being. When the source of happiness is natural acceptance, one becomes self-organized.

Happiness and prosperity are universally acceptable to the self in any human being. Any self desires perpetual or continuous happiness and prosperity. They continuously think about means to fulfil their desire. The needs of a human being are defined on the basis of their desire to be happy and prosperous. They continually act to fulfil their desires as well. However, in most cases, there is always a gap between the state of the self and the state they want their self to be in.

The needs of the self and the needs of the body are characteristically different. The needs of the human body are essential but temporary and limited/quantitative. They are fulfilled by material objects or physical facilities. The needs of the self are continuous, qualitative, and non-finite. Only a few of the needs of the self are fulfilled by materials or physical facilities. Most of the needs of a self are qualitative and are fulfilled by the right feelings.

In many cases, the fulfillment of the desire does not translate into happiness or prosperity. It may only result in temporary pleasure or excitement. Assuming that more of the materials needed for fulfillment of the desire would give more happiness and excess of the material would contribute to prosperity, the self desires and behaves for accumulating more and more material. Irrespective of the quantity of accumulated physical facility, happiness or prosperity is left wanting. This is due to the lack of the right understanding of the source of happiness and prosperity. The misunderstanding leads to the formation of desires enslaved to preconditioning and sensation. The self ASSUMES something to be believed or something someone else does would give them happiness.

The self must have the right understanding that it is the self that needs happiness. One must understand that the need for any physical facility is just to suffice the temporary needs of the body. The understanding that happiness must not be associated with materials and that the source of happiness must be internal would warrant the right behavior with fellow beings and the right work balance with nature.

What is the cause of the unsustainable relationship among human beings and their exploitative relationship with nature? Perhaps a lack of right understanding about a few dozen concepts about life.

This has perhaps occurred as a result of some kind of reducible error with human learning when people started learning concepts from experience without proper self-exploration. Powerful symbolisms proposed as shortcuts to pass on values in an easy manner replaced the actual knowledge about the values. The symbolism, when passed over generations, contributed to information loss and improper interpretations. Gradually people started forming different understandings about the same concepts. Love is a prime example of such a misconception. All religions profess love. It is mistaken for religious zeal. Youngsters and teenagers have a totally different interpretation of love. They misinterpret the pleasurable or exciting infatuation towards another being as love. Parents confuse love with care and guidance. It often manifests as overprotectiveness and overbearing behavior.

Most of us learnt the concept of love either through preconditioning or from the behavior of people using sensation. Very few explore natural acceptance to find the right understanding of care, guidance, and love.

If we were to qualitatively evaluate the performance of human beings with such foundational concepts, what would we find? If we assess human behavior, the efficiency in behaving with appropriate care ranges between 30 to 90 percent. The unsustainable behavior of humanity indicates that the guidance given to the current generation is focused on skill rather than values. A rate of divorce of around 50 percent in marriages indicates that people misinterpret infatuation or physical attraction as love.

Unfortunately, this improper learning is passed on to machines as well. And research often reports models with high performance above 90 percent. Obviously, not all research performs tests on models with exhaustive datasets that include all real-time scenarios.

Perhaps the wrong understanding of concepts that a model learns from the large corpora of data accounts for a part of the irreducible error in ML models - the unknown noise.

Human beings must rise above imaginative realities to find a universally acceptable understanding of existence. We must mend our teaching-learning pedagogy to inculcate self-exploration. We must strive to develop the right understandings, the right relationships, and the right amount of subscription to physical facilities.

We must also evaluate machine learning models qualitatively to ascertain whether an ML model has learnt the right business values. To actually see whether the system proposes sustainable solutions rather than short-term profits.

The learning models perhaps require an overwhelming rework!