An AI-powered Bhagavad Gita project and a question: Who am AI?
What will my friends think of me spouting fake rubbish? Will my twin know my thoughts, memories, things I don't want to talk about?
One wintry evening in Delhi, I received an unusual call from New York. At the other end was a soft, charming voice that belonged to someone who was bringing out the world’s classics using Artificial Intelligence. They’d chosen the Bhagavad Gita from India and wondered if I’d be a guide for this interactive AI book. The voice explained that the AI book would encourage readers to ask questions while reading the text. Answers would come from my cloned voice, based on 15 hours of interviews conducted in advance. At the end of each chapter, the reader would engage in a discussion with my clone.
I was fascinated and perturbed. I wished I’d had such a personal tutor while reading the Gita for the first time. But I was also troubled with questions of identity, selfhood and authenticity. It felt strange having a twin speaking in my voice. Ironically, these are the same questions — who am I? — that the Gita also addresses.
The friendly voice reassured me that their company had already published half a dozen classics. Well-known names like Salman Rushdie and Margaret Atwood had signed up to do future classics. I was ambivalent but I agreed. After hanging up, I was left to deal with my AI monsters.
Nine months later, I’m hard at work on this project. I have three AI interns to assist me: ChatGPT is the most empathetic, Perplexity is best at research, giving links to follow up, and Claude has a literary style. Gemini is a good standby. They are hugely competent and cheerful. ChatGPT can translate a Sanskrit verse into English in 3 seconds, which is 100 times faster than Patrick Olivelle, the great scholar. When I have occasionally pointed out that something is not quite right, they try again, without a hint of complaint or the need for approval. And they keep getting better. While they have taken away the drudgery of research, I’m troubled by my lack of ownership, feeling an alienating distance from the AI output.
Since the invisible labour of research has been eased, I have more time to think, to play thought games: What if the Buddha had been Arjuna’s charioteer? How is it that M K Gandhi, the apostle of non-violence, and Heinrich Himmler, the Nazi apostle of violence, were both inspired by the Gita? Himmler used to train SS officers quoting from the Gita.
I worry about being cloned by a machine. What will my friends think of me spouting fake rubbish? Will my twin know my thoughts, memories, things I don’t want to talk about? What if a friend of mine actually developed a closer relationship with my compassionate clone? I’d feel jealous, cheated, and a fool. It feels eerie having him running around giving talks, making podcasts, racking up thousands of listeners. Worse, what if he cleaned out my bank account, confirming transactions on the phone via voice recognition software?
AI raises fundamental questions about what it is to be human. The only thing I am sure of at this moment are the thoughts and feelings going through my head. If I look into my consciousness, I cannot find the thinker of my thoughts, the feeler of my feelings. Western philosophers and neuroscientists have tried to locate the self but haven’t found it either. The Gita claims that my sense of identity is an illusion, created by the human ego. My true self is the atman, which is identical with the cosmic brahman. Awareness of the oneness of everything is the Gita’s central teaching. Buddha also faced this problem but reached the opposite conclusion. He resigned himself to living with the idea of no-self (anatman).
AI has added a second layer of falsehood. “Who am I?” is a question based on lived experience, not to be answered by a machine. The idea of a “cloned twin” of my thoughts providing answers feels like betrayal. Already shaken by the realisation that my exalted liberal self is a fictional narrator, AI has added a double fiction, leaving me totally befuddled. Mark Twain must have felt this absurdity when he quipped, “There is no difference between fiction and non-fiction — only that fiction has to make sense!”
That AI might become sentient is my other fear. Before it happens, I believe, AI has formidable hurdles to cross. One, the human experience is inherently subjective and cannot be explained in objective terms. When I observe a red apple, it invokes all kinds of feelings of redness inside me. It could be the memory of blood oozing out of my brother’s leg in an accident; or the red cheeks of Renoir’s young girl in his painting. Philosophers call this subjective content “qualia”. The question is how AI will duplicate qualia, the private, unspoken product of the subjective consciousness of eight billion humans?
Another obstacle is to replicate my brain’s biological neuron-specific physiology in my subjective experience. AI may copy the objective red colour — its wavelength of 650 nanometres — but it cannot access my subjective redness that emerges from my brain’s 86 billion neurons and 500 trillion synapses. Both these hurdles lead to a third one. Essential to human consciousness is spontaneity, a freedom to choose my goals. That autonomy, the Gita tells me, comes from the life force, atman. AI today can follow commands based on algorithms and data. It has far to go to achieve self-awareness.
While I am charmed by the wonderful possibilities of AI in teaching classics to the world, I am deeply concerned about its downside. Re-reading the Gita, however, has made me rethink my own values. In the “real world” of work, I was valued for being sharp and acquisitive. What I admired, though, were kindness and generosity, traits of failure in the “real world”. What is at stake is our conception of success. The present crisis in public life around the world may lie in this. There are parallels between the Gita’s lament and the things we might say about our leaders today.
Das has interrogated the Mahabharata in his book, The Difficulty of Being Good
© The Indian Express Pvt Ltd