Vedic Literature and AGI
The Vedic literature is vast and voluminous, and the Susiddha AI project will use it to build the Virtual Veda Vyasa (V3) artificial general intelligence (AGI) system. The previous chapter on the Sanskrit language spoke of the Vedic core, where the AGI system will do its core thinking, understanding, and reasoning. Vedic literature is central to the operation of the core.
The Vedic literature covers a wide range of subjects and fields, including spiritual, philosophical, scientific, medical, legal, political, and economic. It covers the entire range of life. One of the important principles of Vedic literature is the “four aims of human life”; in Sanskrit these are: kāma (pleasure, entertainment), artha (livelihood, wealth), dharma (duty, good conduct, virtue), and moksha (enlightenment, liberation). The Vedic literature facilitates the attainment of these four aims, and covers everything about human existence.
This chapter is not an overview of the Vedic literature, and can only touch on a few aspects of it. A subsequent chapter on Cognitive Architecture will discuss the use of some specific branches of the Vedic literature. [1]
Vedic literature has its origins in an oral tradition, of which the main surviving work is the Rig Veda. The Rig Veda is the anchor of the class of works called “shruti”, which means “that which is heard, or is to be heard”. Shruti will be discussed in a later chapter where we discuss how Rig Veda and other shruti will be learned by the V3 AGI system, not by reading text, but by listening to oral recitation, and processing that into deep neural networks. Because shruti was originally transmitted orally (without any written text), and it involves nuances of tone and accent that cannot be accounted for by a written text, the true meaning of Rig Veda and other works of shruti can only be grasped when learned orally (and “aurally”).
Shruti is composed in the older Vedic language (discussed in the previous chapter on Sanskrit. This includes some of the works that are classified as Samhitas, Brahmanas, Aranyakas, and Sutras.
Beyond shruti, the later works of Vedic literature are composed in the classical Sanskrit, and most of these works were written out, although many of them stem from lost works of the oral tradition.
Works written in classical Sanskrit do not have markings for tone and accent. (However, any work written in a Sanskrit meter has implicit accents.) Thus these works can be learned and processed via textual methods, such as those used in natural language processing (NLP) and computational linguistics. (However much can be gained by processing them “orally” as specified above for shruti.)
Because the Vedic literature is so vast, decisions will have to be made regarding what works to include in the core, and what weight to give each work. How to make such decisions will be an item of the research agenda of the Susiddha project. Also, later chapters will explain the use of probabilistic logic and uncertain inference which will play a part in all knowledge that is stored in the system.
In creating the V3 AGI system, the Vedic literature will be used extensively to guide the design. However, we will avoid reinventing the wheel. Vedic knowledge is ancient, but we won’t ignore valid knowledge (e.g. from cognitive science) that has been discovered since Vedic times. We are also free to use and adapt components of other open-source AGI systems that fit within the guidance provided by the Vedic literature.
In the following chapter on Cognitive Architecture we will highlight some of the aspects of cognitive science found in the Vedic literature and discuss how they can be used to guide the development of the V3 AGI system.
Notes and References
- The term “Vedic literature” is always used in this website, as opposed to the term “Hindu literature”. The term “Hindu” was coined by foreign invaders, and popularized by the British who were the last foreigners to invade India. The term “Vedic” is an adjective of the Sanskrit “veda” which means “total knowledge”. All Vedic literature is written in Sanskrit.
Susiddha AI Project 