Can Machines take over Humans?

As Artificial Intelligence (AI) is getting more and more sophisticated, many jobs requiring decision making in ambiguous conditions are now possible by algorithms. Hence, artificial bots are replacing human workers. This is dramatically reducing the number of jobs in which humans are required. It is raising a fundamental question, what is the value a human can contribute which a machine cannot. Future jobs will be decided by that value.

The above is a layman's perception of the situation. A fundamental question arises, if technology advances as much as it could, how powerful can machines get relative to what a human can achieve. To understand this, we need to grasp what is the core capability that is leading to the ability to automate. 

Initial attempts of Artificial Intelligence tried to mimic the inferencing capabilities of humans. This failed miserably. The real breakthrough came with machine learning and its ability to identify patterns in past data. Initially these patterns were primitive. But with the advent of Convoluted Neural Networks (CNN) a.k.a. Deep Learning, the patterns became more sophisticated. Now we can identify recurrence of even patterns of inference. The fundamental capability is this: by capturing human behavior in large quantities a machine can identify recurrent behavior with a high accuracy. As a result a machine can learn from past patterns. The flip side is that the machine can only repeat the past but not compose new inferences. This limits creativity. So what humans can add is creative solutions that are not repetitions of past patterns.

To the extent a human action is a repetition from memory, whether individual or collective, a machine can replace. This forces humans to develop and rely more on intuition for decision making to stay ahead of machines.

According to Vedic wisdom a human is a collection of multiple nested personalities or identities called the koshas. The human experiences the world via these identities and stores the outcomes in a common memory called the chitta. He gives value to an experience by comparing it with stored experiences of various koshas in the chitta. His  reactions to the incoming shocks of the world are predominantly conditioned by the stored experiences. These stored experiences are called samskaras. Examples of samskaras: Muscle memory at the physical level, urge to defend ego at pranamaya kosha, instinct to protect family, prejudice to one's own ideas, and affinity to dharma. Human behavior is predictable to the extent it is determined by samskaras. This is what can be accurately modeled by machine learning.

What distinguishes a human from a machine is his sense of self, its experience and the joy resulting from it. The machine does not have any of these. All it has is the memory and the ability to search it for a  match. To the extent human inference of new facts from existing data follows a pattern, even a machine can mimic that. A machine cannot feel, but can process feelings as accurately as a human. This is because humans are predominantly memory-driven i.e. samskara-driven or habit-driven. 

True liberation (Moksha) is therefore liberation from bondage to the past. AI is hastening humanity towards liberation.