The ‘golden’ age of jyotiḥśāstras (treatises about astronomy) occurred between the third and ninth centuries CE, but only a few prominent texts have survived into the modern day: this is largely because they made their way into practice Jyotiṣa (astronomy) was and continues to be an important part of South Asian Culture—for assigning appropriate names for children according to birth signs and times, selecting life partners, finding auspicious dates for events such as marriage, moving house etc.
But with these texts being so old, and often written in Sanskrit, they have always required an intermediary, in this case an astrologer, to calculate, authenticate and interpret the various signs, charts, tables and stars. Within the last few years, we have seen the rise of Generative Pre-training Transformers (GPTs) in modern society. In this wave, applications such as KundaliGPT and VedicAstroGPT make use of the technology and artificial intelligence to give astrological advice, circumventing the need to involve a human practitioner.
In this paper, I shall discuss how some of these GPTs work, but more importantly I will explore the implications of such technology and its relationship with Sanskrit śāstra, experts and society, including looking at potential benefits, pitfalls and tentative ideas on what the future of astrology and ritual are in an AI, technology-centric future, especially as it moves away from the human and personal aspect of intermediaries and experts, based on my interactions with the creators and users of these GPTs, as well as traditional jyotiḥ practitioners and their customers.
I shall do this with a confrontational ethnographic method, where I show traditional astrologers in various cities in Gujarat, West India, these astrology GPTs and then ask them about their experiences, thoughts, fears and worries. I also do the same with users of both traditional astrologers, as well as users of these astrology GPTs. Additionally, I interview and discuss the interaction between GPTs and 'traditional' forms of knowledge in jyotiḥ with one creator of such a GPT.
In this way, I shall try and determine how AI and machine learning models are intervening upon traditional knowledge systems, and how they are viewed, challenged, and even integrated into practices of traditional knowledge.
Likewise, I use the intervention of GPTs and AI to challenge ideas of where "efficacy" lies in the astrological process (whether it lies in the "science" of astrology itself, or in the astrologer, and if the latter, whether this efficacy can be supplemented by technology or whether it is reliant on a human practitioner), thus contributing to ritual studies and studies of traditional knowledge.
The golden age of jyotiḥśāstra (astronomical treatises) spanned the third to ninth centuries CE, with select texts surviving into modernity. Jyotiś (astronomy) remains integral to South Asian culture, determining auspicious times for lifecourse events and guiding personal decisions. Traditionally, these Sanskrit texts required astrologers as intermediaries for calculation and interpretation. Recently, AI applications like KundaliGPT and VedicAstroGPT have emerged, offering astrological guidance through generative pre-trained transformer technology. This paper examines these AI systems' functioning and explores their broader implications on the relationship between Sanskrit śāstra, expert knowledge, and society. The analysis considers potential benefits and challenges as astrological practice transitions from human intermediaries to technological interfaces. By examining this intersection of ancient knowledge systems and modern AI, this study contributes to understanding how traditional practices adapt in technology-driven environments and questions what may be gained or lost when ritual practices become increasingly automated.