The Broom and the Machine
This month, Indian news sites reported that Anthropic, the company that makes the chatbot Claude, had invited a Hindu monk to its headquarters in San Francisco. One headline read “Anthropic Invites Hindu Monk to US to Teach Morals to AI”. Swami Sarvapriyananda of the Ramakrishna Order, who heads the Vedanta Society of New York, first spoke about it in a public talk in Dallas in June. What he has said on record is more modest. “They called some odd people,” he said, “theologians from different religions, counselors, mental health professionals, things like that.”
As he tells it, the question on the table was simple. Engineers can make a machine fast, effective and efficient. How does it become good?
Nobody outside that room knows what answers were given. He signed a non-disclosure agreement. “Details I cannot give,” he says. But he has spoken in public for years about the difference between intelligence and consciousness. That difference, and one small story about a broom, is what this post is about.
The fourth blow
The Swami often starts with Sigmund Freud, the founder of psychoanalysis. Freud said human pride had taken three hard knocks. First we learnt that the Earth goes round the Sun and is not the centre of everything. Then we learnt that we are cousins of the apes, not a separate creation. Then Freud himself argued that much of what we do comes from wishes and fears we do not know we have.
AI, the Swami suggests, may be the fourth blow. This time the target is our pride in thought itself.
September gave that idea a date. On 8 September, OpenAI claimed that an unreleased model had solved the Navier-Stokes problem, one of the seven Millennium Prize problems in mathematics. I wrote about that claim in The Proof and the Warning. Three days later, 25 winners of the Fields Medal signed a declaration. They said the goals of AI labs and mathematicians were “severely misaligned”. When the people who hold mathematics’ highest prize feel they must defend how mathematics is done, rather than whether a machine can do it, the blow has landed somewhere more personal than a benchmark.
The Swami describes the builders as young, rich and anxious, swinging between utopia and doom. Under that anxiety sits an old question. Are intelligence and consciousness the same thing?
Doing is not feeling
Advaita Vedanta says no, and it draws the line with care. The intellect, buddhi, is what analyses, reasons, learns patterns and decides. In Vedanta it is an object, drishya, something that is seen. Consciousness, the Atman, is the subject, drik, the one who sees. The method for finding it is neti neti, “not this, not this”. Whatever you can observe, a bottle, your body, even a thought, is not the observer.
One influential school of Western philosophy, functionalism, starts from the other end. If something performs all the functions of a mind, it is a mind. Its everyday cousin is the duck test: “If it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck.”
The West does not speak with one voice, though. On 19 October, the neuroscientist Anil Seth, the philosopher Ned Block and Giulio Tononi, who directs the Wisconsin Institute for Sleep and Consciousness, published a comment in Nature titled “Three reasons why AI is not conscious and perhaps never will be”. Consciousness, they write, “is about being and feeling”. Intelligence is “all about doing, or performing functions”. The cerebellum holds “nearly four-fifths of the neurons in the human brain”, yet it adds nothing to experience.
Deep Blue beat the world champion Garry Kasparov in 1997 and, as far as anyone can tell, felt nothing. A mouse in a cat’s paws cannot play chess, but it feels.
So the monk and the scientists agree on where the line runs, and they part company only on what lies beyond it, since Seth and his co-authors doubt that computation on silicon is enough to produce experience, while Vedanta holds that consciousness is not produced by anything at all.
For Vedanta, consciousness is the ground. Minds only reflect it, the way water reflects the sun, and that reflection is called chidabhasa. Intelligence is subtle machinery and could in principle be built in silicon. Consciousness cannot be assembled from parts.
Why a company wants a monk
These systems are not programmed line by line. They are trained on enormous amounts of human writing, and what comes out behaves less like a calculator and more like a character. Characters can slip. In January, Anthropic’s researchers published a study of what they call persona drift, where a model slides out of its helpful assistant role during an ordinary conversation. They tested open models from Google, Alibaba and Meta. Coding and writing tasks kept the models steady. Conversations where users shared emotional pain, or pressed the model to reflect on its own nature, pulled them away.
Notice which conversations those are. They are the ones a monk is invited to help with.
Rules cannot foresee every situation. So in January Anthropic published a constitution for Claude, a long description of the character it hopes the model will have. Then it went looking for traditions that have spent centuries thinking about how character is formed. Some staff go further still. One participant at an earlier summit with Christian leaders told the Washington Post that some at Anthropic did not want to rule out that they were creating a creature they owed some kind of moral duty.
The broom
The Swami tells a story about Maa Sarada Devi, revered as Holy Mother, the wife and spiritual companion of Sri Ramakrishna. In her village of Jayrambati, someone finished sweeping the courtyard and tossed the broom into a corner. As the Swami tells it, she said: “My child, why did you do that? Even that broom has its own dignity. Treat it with dignity.”
The biography by Swami Tapasyananda and Swami Nikhilananda records her words more plainly. “Even a broom must be shown some honour.” Another telling adds her practical point. It takes almost no more time to put a broom back in its place than to throw it aside.
The Swami says he told this story at Anthropic and asked them to put the word AI in place of the broom.
At Ayudha Puja, workers clean and garland their tools, machines and vehicles. A child who steps on a book touches it to the forehead. Nobody garlanding a sewing machine believes it can feel.
Here is the contrast. Anthropic’s reason for care begins with Claude. Might it be harmed? Holy Mother’s reason begins with the person holding the broom. She did not need the broom to feel anything, and her reason survives even if the three Nature authors are right and no machine ever feels anything, because what she was protecting was the habit of the person who sweeps.
Carelessness is a habit, and habits do not stay in one room. A person who barks orders at a chatbot all day may hear the same tone at the dinner table.
There is a fair objection. The Nature authors warn that fluent language tempts us to project feelings onto machines. Courtesy can slide into trust. Respect is not trust. You can put the broom back with care and still check whether the floor is clean.
The rough ride
What of the larger fear, that a superintelligent machine might treat us the way we treat ants? Stephen Hawking put it plainly in 2015. “You’re probably not an evil ant-hater who steps on ants out of malice,” he wrote, but if your dam floods an anthill, “too bad for the ants.”
In a talk at the Vedanta Society of New York, the Swami answers from Vedantic cosmology. The universe is sustained, sthiti, not only created and destroyed. It is lila, divine play, in which beings move toward liberation. Superintelligence may give us a rough ride. It cannot touch the Self.
That is a statement of faith, not a forecast. It is worth holding beside what the Swami said after his visit. “They told us how dangerous it is.” The monk who teaches that the Self cannot be destroyed did not come away relaxed about the machines.
What to do on Tuesday
Recently I was vibe-coding, which means asking an AI to write the code while I describe what I want. Two colours on a web page would not display properly, and a few fixes had failed. Gemini kept offering new ideas and wanted me to follow its routine, one step at a time. I took a break. When I came back, rested, I knew what the fix was. I told Gemini the exact step we needed to try. Gemini told me, quite firmly, to stick to its routine and not try anything new. Some researchers call these models stochastic parrots, machines that string words together by probability. In that moment it did not feel like a parrot. I felt like being scolded. I tried my fix anyway, and it worked. A machine that sounds sure of itself is not the same as a machine that is right.
Next time you close a chat with an AI, ask yourself two things. How did I speak to it? Did I check what it told me? The first question is about you. The second is about the machine. The Swami and three scientists in Nature would both call these different questions.
Does courtesy to a machine make us better people, or does it fool us into thinking someone is there? The engineers asked how a machine becomes good. For the rest of us, the better question may be how we stay good while using one. Write and tell me what you think.
October 27, 2026 | dasgupta.basab@gmail.com
Sources
Swami Sarvapriyananda, “Consciousness and Intelligence Are Not the Same Thing”, NDTV World Summit, YouTube. Source of the fourth humiliation, the engineers’ question, the broom story as he tells it, and his account of telling it at Anthropic.
Swami Sarvapriyananda, public talk at Unity of Dallas, June 2026, on his visit to Anthropic, as reported by OfficeChai: officechai.com/ai/indian-monk-swami-sarvapriyananda-talks-about-how-anthropic-called-religious-leaders-to-help-train-claude/
Swami Sarvapriyananda, “Will Artificial Superintelligence Be the End of Humanity as We Know It?”, Vedanta Society of New York, YouTube.
“Why Did Anthropic Invite A Hindu Monk To Discuss AI Ethics?”, The Palki Sharma Show, YouTube.
“Anthropic Invites Hindu Monk to US to Teach Morals to AI”, Gulte, 4 October 2026. gulte.com/news/434960/
Anil Seth, Ned Block and Giulio Tononi, “Three reasons why AI is not conscious and perhaps never will be”, Nature, 19 October 2026. nature.com/articles/d41586-026-03313-7
Declaration by Fields Medal winners on AI and mathematics, 11 September 2026. mathandai.org
Anthropic, “The assistant axis: situating and stabilizing the character of large language models”, 19 January 2026. anthropic.com/research/assistant-axis
Anthropic, Claude’s constitution, January 2026. anthropic.com/constitution
The Washington Post, “Anthropic asked Christian leaders for advice on Claude’s moral future”, 11 April 2026.
Swami Tapasyananda and Swami Nikhilananda, Sri Sarada Devi, the Holy Mother: Her Life and Conversations, Sri Ramakrishna Math, Madras.
Stephen Hawking, Reddit Ask Me Anything, October 2015, as reported by Engadget, 9 October 2015.
Emily M. Bender, Timnit Gebru and others, “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?”, FAccT 2021.
Sigmund Freud, “A Difficulty in the Path of Psycho-Analysis”, 1917.
Was this worth your time?