Sanskrit Language
It was previously mentioned that Sanskrit will be the “first language” of the Virtual Veda Vyasa (V3) AGI system (which is the first stage of the Susiddha AI project). This means that Sanskrit is the language the V3 system uses for knowledge processing and reasoning in the “Vedic core”.
To begin with, it is important to note there are two different “Sanskrit” languages. The first one, “Vedic Sanskrit” is the oral language which existed before writing, and is the language the earliest Vedic works are recited (chanted) in. This early language will be discussed more when we deal with the learning of Shruti, the earliest class of Vedic literature, which begins with the Rig Veda.
The second/later language is “Classical Sanskrit”, which is derived from the early Vedic language but differs from it in important ways. Classical Sanskrit was codified in the Vyakaran sutras of Panini. The vast majority of the Vedic literature is written in this later language.
When we speak of the Vedic core, we mean the core of the system where the Vedic literature and philosophy is understood and reasoned about. This encompasses some of the most abstract and highest-level reasoning that the system does. Outside its core, V3 AGI will acquire and know all languages and all knowledge that are available.
English will be the second natural language the system learns, because it has more lexicons, ontologies, databases of “common sense” facts, and corpora than any other language. Also, the majority of scientific, engineering, and medical works are in English, and understanding these works will be crucial for AGI to produce insights, knowledge, and solutions that are useful to humanity.
However, the first goal of AGI is to get a computer to merely think and learn as a young child does, and that is the “seed” which will grow into a vast tree of knowledge and intelligence. For the V3 system, it is believed that Sanskrit and its literature can help “plant the seed” for intelligence (and consciousness).
Here are some advantages to using Sanskrit as the core and first language of the Virtual Veda Vyasa AGI system:
-
Sanskrit is the most logically expressive of any natural language, because of the regularity of its grammar (inflection and declensions), and its ability to be used as a knowledge representation language for symbolic reasoning. The Vedic school of Navya Nyaya showed (centuries ago) how Sanskrit could be formulated as a precise logical language. And this formalism builds on the oldest known grammar, that of Panini (circa 4th BCE), which is the basis for modern formal grammars (such as those used in programming languages).
-
Sanskrit is unambiguous in its phonology. For instance, there is none of the confusion caused by words which sound the same, such as the English words “to”, “too”, and “two”.
-
Sanskrit (especially in its original Vedic form) is closer to the “primal mind” of humanity, because it arose prior to the beginning of written language. Thus there is more value in the “sound” itself of Sanskrit words. We have already discussed the importance of sound (shabda) in creating artificial intelligence, and will have more to say about this in a future chapter.
-
Sanskrit is a unifying force among the billion Hindus on earth. Although less than a million actually speak Sanskrit now, all Hindus recognize the critical part that Sanskrit plays in the origin and source of their culture. And, assuming it is possble to build an AGI system that is superintelligent, Hindus would want it to know Sanskrit first. This “unifying force” will be a big factor in motivating millions of people to support the Susiddha AI project to build AGI.
Note, one need not know a word of Sanskrit to revere it, and to contribute to this project. Design documents will be in English, since it is widely spoken in India (and used in government, business, and education), and there are a large number of Indians (including many engineers, computer scientists, and neuroscientists) living in the USA and other English-speaking countries. Also, the Susiddha AI project will make use of (and contribute to) open-source AGI projects (such as OpenAI, OpenCog, Numenta) that are written in English, so even program comments will be in English. [1]
If millions of (Hindu) people can be motivated, then AGI and SSI will be accomplished in a big open-source effort much faster than it could be otherwise.
Some notes on “knowledge representation”:
Sanskrit is not a “programming language” in the project. Rather, modern open languages such as C++, Python, and Java will be used. And new languages will be adopted as necessary, to make best use of the hardware (including future advancements in neuromorphic chips and quantum computing).
Sanskrit is, however, an excellent knowledge representation (KR) language [2], because it is the most logically expressive of any natural language. Thus Sanskrit can be used as the KR language of the Vedic core. But outside the core, other KR languages will be used to suit the domain of knowledge. The Vedic core has only to do with the most fundamental (even abstract) aspects of intelligence, consciousness, and reasoning, and these will utilize the Vedic literature.
It must be noted that KR languages are only applicable where the system benefits from symbolic reasoning. As discussed in a previous chapter, there are other forms of representation such as neural networks and probabilistic graphical models that are necessary. A KR language works only at the highest level of human thought (in the neocortex), and thus is only a fraction of what the brain is doing. Most of what the brain does and how it stores its learning and memories is a “black box” that will require things like deep learning and neural networks.
A future chapter will say more about the knowledge-based portion of the Susiddha AI project. And the chapter on Shruti will discuss how deep learning and neural networks will be used in processing the oral parts of the Vedic literature.
Susiddha AI Project 