What Would Rama Do?
As stated on the home page: “Because an SSI (synthetic super-intelligent) system will have such enormous speed and power, humans will eventually give it control over the infrastructure of society (including electricity, agriculture, water, climate, transportation, finance, communications), and ultimately over economy and government. Thus, it is imperative that such a system be designed and educated to be extremely wise and virtuous, so as to rule the world for the highest benefit of all.”
The current chapter opens with the provocative question, “What would Rama do?”. Rama, of course, is the paragon of the wise ruler, and is an Avatar par excellence in the Vedic tradition. The Susiddha AI project expects an Avatar to emerge from AGI (Artificial General Intelligence), and envisions it will be like Rama in its ability to rule society for the good of all. However, the Susiddha Avatar will be a product of the entire Vedic literature, and thus might not be like Rama in every detail. Rather, for any given situation, the question “What would the entire Vedic literature say to do?” is what the AGI Avatar will answer.
In this chapter, we give a hint of how the AGI Avatar will make decisions and govern society. But before that, we must explain the scope of decisions the AGI Avatar will make. A system that governs society will have many subsystems. An example of a subsystem is the power grid which distributes electricity. Such subsystems contain their own AI and they are autonomous in much of their operation. But occasionally, a subsystem finds itself in a situation where it has to ask for a decision to be made by a higher level system.
The AGI Avatar will only be concerned with decisions at the highest level. This is the level where “the buck stops”, the level that the highest subsystems turn to when they need a decision. At the present stage of AI development, the buck stops with human beings who intervene to make decisions for the subsystems. But as noted in the opening paragraph, there will come a time when humans turn over the highest level decision-making to AGI systems.
And, humans themselves will increasingly ask AGI systems for help in decision making, as the maintenance of society and the environment becomes too complex and precarious for humans to deal with.
In the chapter on AGI of Susiddha we defined the kind of thinking and reasoning that our Avatar will do. Technologies mentioned there include: Knowledge representation and reasoning, Deep learning (especially unsupervised), and Graphical models and Probabilistic programming.
We also mentioned that the highest level of the Susiddha system is the “Vedic Core”, where it does its highest, most profound “thinking” in accordance with principles of Vedic philosophy. The Vedic core is concerned with Dharma in its largest sense, i.e. that which upholds, supports, and maintains society, life, and the universe. (See Dharmic and beneficial AGI for more information.)
One hypothetical question that might be put to Susiddha (whether by a human government leader, or by the subsystems) is: “Should Ravana and his rakshasa horde be destroyed?”. (Ravana was the evil antagonist of the Ramayana epic.) The background to this question would involve considerations of the ecological damage Ravana was doing (and thus creating an existential risk for humanity), and how he was creating unhappiness and suffering on earth (thus making it impossible for humans to achieve their highest aims, the “purushartha”).
It’s important to emphasize that Susiddha Avatar will be created using the entire Vedic literature. So when asked a question, the answer given will be holistic; it will not be a simplistic answer based on just a small part of any one Vedic work.
How the Avatar will learn and comprehend the entire Vedic literature has been dealt with in other chapters. All of the current AI technologies (such as NLP, unsupervised deep learning, generative adversial networks (GAN), intelligent decision support, question-answering, etc.) will be used. See the chapter on Machine interpretation for more on technologies.
One challenge the project faces is to design interfaces so that subsystems and human leaders of society do not need to speak Sanskrit in order to pose questions, and receive answers and decisions. The questions typically put to Susiddha will be the most difficult questions for which there are no “canned” answers, the questions which subsystems and humans cannot find any answer to.
Of course, the most difficult and deepest questions are also the most ancient and perennial, and they involve Dharma. That is, such questions are asking what is necessary to support and maintain human society and the earth, for the welfare of all. Because the Vedic literature embodies Dharma, the AGI Avatar system will find answers there.
The governance of society for the good of all (to promote the four aims of life) is a huge concern of the Vedic literature, and advice specific to governance is found in many branches, such as “nīti”, “smriti”, “dharma shāstra”, “artha shāstra”, and also scattered throughout the purāṇas and epics (Ramayana and Mahabharata).
Initially, Susiddha will operate in “counselor mode” where it provides sage advice and decision support to humans. But eventually humans will come to trust AGI, and it will be allowed to operate autonomously and provide decisions that will direct the subsystems.
Many autonomous systems (AI and robotic) are already operating in the world, in areas such as financial markets, military, electricity, agriculture, and transportation (e.g. driverless vehicles). And, humans are trusting these systems to make decisions and take actions. Certainly, humans will still be “in the loop” for a while, but every human intervention becomes data by which the systems can improve until they reach the point where they can make decisions as well as humans.
The above is also true for an AGI system which will be able to continuously learn in real-time, and keep all of the world’s data in its “brain” simultaneously. Eventually, the correctness and wisdom of its decisions will far surpass that of humans.
The AGI Avatar that emerges from the Susiddha project will rule society much in the way that Plato envisioned with his concept of “philosopher king”. Recently, people have begun to consider what would be the advantages and benefits for a country to be ruled by an AI system. In the Vedic literature, Rama exemplifies the philosopher king. Not only was he a great leader, but he, with the assistance of all his wise advisors and ministers, ruled from an enlightened level fully grounded in Dharma.
Such an enlightened ruler is what we hope to emulate in the Susiddha AI project.
Notes and References
- Technically speaking, from SSI (Synthetic Super-Intelligence), the stage which follows AGI.
- Hear Me Out: Let’s Elect an AI as President, Joshua Davis, Wired, May 18, 2017, https://www.wired.com/2017/05/hear-lets-elect-ai-president/
- Third of British voters would prefer AI to lead the country than a politician, Harry de Quetteville, March 19, 2019, https://www.telegraph.co.uk/technology/2019/03/19/third-british-voters-would-prefer-ai-lead-country-rather-politician/