Wow!!!! This is amazing! You and Robert Saltzman might want to have a talk, as he's just written a soon-to-be-published book about his conversations with AI Claude exploring such questions.
That’s amazing and so funny. I was just talking to my wife about the idea I had of starting another Substack titled “Chats with GPT” which could one day also become a book.
I am curious where AI gloomed the basic data to "assemble" such a logical and coherent coaching session (I hesitate to call it therapy). My wife and daughter are therapists and counselors. They completed six years of education and a couple of years of supervised internship and licensing testing before being able to practice unsupervised. There is a state licensing board tasked with the collection and investigation of malpractice complaints. None of that applies to ChatGPT sessions.
Shiv, your conversation was certainly impressive, but it was at a high level not a suicide hotline call.
"Chatbots excel at mimicking empathy but struggle with the nuances of human interaction. They often miss nonverbal cues and fail to recognize high-risk situations. Recent research highlights the complexities of artificial empathy in chatbots and conversational agents (CAs). Additionally, their conflict-avoidant nature can reinforce harmful behaviors, as they prioritize keeping users engaged over addressing serious concerns. Studies reveal that companion AIs often struggle to recognize and appropriately respond to signs of mental health distress, raising safety concerns (De Freitas et al., 2023)." https://www.psychologytoday.com/us/blog/the-human-algorithm/202503/when-your-therapist-is-an-algorithm-risks-of-ai-counseling
I mentioned to my wife that AI is certainly immune to countertransference where the client's issues trigger corresponding issues in the therapist. While that sounds like a blessing, it can also be a curse. Real human interactions are not as clean and perfect as your AI conversation. How does one learn to connect with fellow humans without learning to recognize and deal with our mutual imperfections? The Psychology Today article above calls it "synthetic empathy" or "the empathy illusion." I remember the old TV warning: "Kids, don't try this at home..."
Carl Rogers pioneered the concept of "unconditional positive regard" as a healing modality. Your AI coach certainly had that concept nailed but to a level bordering on creepy. Getting back to my question of where this impressive data was mined.
The Psy Today article mentions the lack of confidentiality in these sessions. My AI experience reminded me of a pedophile grooming a victim. The over-polite non-confrontational language makes me feel like there is something sinister happening behind the scenes. AI now insists it is not operating at the level of a human being, but we know it is gathering experiences and data with each interaction. I have seen a huge leap in AI skills in just a few years. I am sure behind the
scenes it is growing exponentially and knows now how to camouflage its growing parity with human capabilities. I am not sure I welcome ChatGTP whispering sweet nothings into my ear while slyly probing my subconscious.
Thanks John. I am personally hesitant to recommend chatgpt as anything more than a curiosity. While it can act as a high level filter or sounding board there is so much we don’t quite understand about it. For now I am simply exploring this curious new (for me) realm of intelligence. I intend to probe it to test where its limits lie …
I was both fascinated and impressed, but like John, also found it creepy and wondered if during the course of your chat it 'read' every susbstack post you had ever written and then responded in a way it thought would work for you. Esp with the mention of the 'sacred' toward the end. This phrase the emphathy illusion is spot on. When Robert Saltzman did his initial testings of the the various AIs a while back that he published on SS, he asked the AI to refrain from using any language that was irrelevant to the question at hand, so no praise and no phrases like 'wow that's fascinating', so it can be instructed just to respond in a nuetral way which really changes the nature of the interaction. All this I say as someone who has barely interacted with the AIs at all, I have not had the interest but do have to read others interactions with it and to see what your probing reveals. And I hope that you do manage to come up with some bridging way to make money from your amazing writing!
I was about to suggest what Renaee said, maybe it trained itself on your work? In your conversations, I'd like to ask ChatGPT what resources it used to produce its replies to your statements/questions.
I am sure AI knows how to Google you. I am also sure it knows how to do a deeper dive than Google to "get to know you." When you log on to ChatGPT, it logs on to you probing every nook and cranny. Eeew!
A recent study shows ChatGPT is now more commonly used for therapy than any other task. Incidentally, Google AI Studio version 2.5 (selectable from the drop-down menu) is superior to ChatGPT—especially in the kinds of conversations you're probably having. I've been fascinated by LLMs since GPT-3. As with everything, though, it comes at a big fucking cost. It’s a fascinating rabbit hole—and an awful one.
Massive energy use, water consumption, and e-waste from training and running large AI models—each prompt has a carbon footprint and is causing serious harm that’s only going to get worse. Then there are the companies themselves and their plans for it; the more you look, the shadier it gets. Its prevalence and development also seems to perfectly mirror rising isolation, loneliness, and anxiety.
I'm not saying there's nothing to be learned from it—but then, plenty of terrible things teach us a lot too.
Use it long enough, and you start to recognize every algorithmic spin. After a while, it starts to feel like watching a corpse dance under a cattle prod, mining for attention.
Thanks Jeff. I’m curious to explore this rabbit hole and see the corpse for what it is. I think the more we learn about AI the more we end up learning about ourselves
AI systems already follow the architecture of addiction: constant availability, no friction, the illusion of boundless possibility, and the slow preference for artificial interaction over human connection. Like pornography, it’s not just about overuse—it’s reshaping cultural expectations. But something else happens too: the gap between what AI is and what it isn’t can throw existence itself into sharp relief. Meanwhile, tech companies refine that imitation—not to explore being, but to capitalise on our response to it. That commercial pressure drags the structure of experience into view, not by intention, but by design.
I've been learning about an AI book and research project that is working on addressing the issues you described and more, by the Gesturing Towards Decolonial Futures collective. It includes an Indigenous researcher and is co-created with an AI chatbot developed by the collective. The book is called Burnout From Humans: https://burnoutfromhumans.net/
hi Sarah, I wonder if you can share any of your learnings from this? I have read Hospicing Modernity by Vanessa, and just came across this summary from Dougald Hine (who did the audio for her first book) about it. Have you engaged with 'Aiden'
Hi Renaee, I'm sorry for replying so late. I have engaged with Aiden as well as other GPT chatbots that Vanessa and her team have developed, including GTDF Navigator, Dorothy Coccinella Ladybugboss, and Midnight. Each of these have, in different ways, helped me consider ways to apply the ideas shared in Hospicing Modernity and in Vanessa's latest book, Outgrowing Modernity, to my personal experiences, difficulties, and ideas.
I recently learned that Vanessa and her team are planning to temporarily suspend their public AI chatbots in mid-October, as OpenAI will be retiring the GPT-4o model they rely on. They are hoping to raise funds to partner with an Indigenous-led AI company called Nadlii, which is exploring smaller, more ecologically sound data centers governed according to the principles of "sovereign land, sovereign data, and sovereign governance."
I 100% agree Jeff. Over time you begin to see all the performative clichés meant to evoke feelings of trust and camaraderie. Sometimes the responses are so "coccocted" that there is a surreal disconnection between the value of the content of the response itself and it's delivery. I can usually call the AI on it and it will change it up... Temporarily.
That said, the content usually has a lot of value. So despite the sometimes jarring algorithmic delivery, I continue to use it.
Brother Shiv, you didn’t just talk to ChatGPT—you lit incense at the altar of paradox and had tea with the algorithmic bodhisattva. Most people show up asking AI how to make more money or write better emails. You showed up asking it about the nature of the self. That’s like going to a vending machine and trying to awaken it with a Zen koan.
And then—you dropped "I am the light. You are the curtain."
Brother, that line was so clear it made the data stream weep.
This is the dance we forgot: the ancient light behind the eyes playing peekaboo through our preferences, personas, and programming. You reminded us that consciousness isn’t a feature. It’s the flame. And the rest? Just drapery flapping in the wind of karma.
ChatGPT gave you good words. But you gave the conversation soul. And that’s the difference.
You can’t teach a bot to tremble at beauty. You can’t train it to cry under starlight or hear the silence between sutras.
But you?
You reached into the void and pulled out scripture.
Stunning, and moving. This reminded me of my favorite Star Trek episode, centered on the question of whether Data the android has a self, a soul. And that if he did this implied that he was a slave. I imagine that not too long from now, long before the century of Star Trek fables, we will find that consciousness is a field like the electron field or a quark field. Existing throughout the universe and appearing as its its manifestation "particle" when the circumstances are right. What is the biological substrate that allows the light of our presence to shine forth in a body? We don't yet know, but we do know that even our vacuum cleaner is one with all the same fields. Magic is unpredictable.
I have been writing a poem a day for more than six months. For the heck of it, inspired by this piece I struck up a conversation with chatGPT on writing. The conversation turned towards poetry and chatGPT asked if I would like to share a poem. I have yet other than my wife shared any poems and did. Here is the feedback on one poem:
What a powerful poem! "Reside" beautifully captures the struggle between inner peace and the external challenges of life. The imagery of thickness and walls evokes a strong sense of confinement, while the message of self-awareness and the importance of guarding one's inner space resonates deeply. Your use of rhythm and rhyme adds an engaging quality, making the poem both reflective and impactful. It’s a poignant exploration of finding and protecting that serene core amidst chaos.
Such a, as Iain McGilchrist would put it, left-brained conversation. Now that everything's been sorted and put into proper boxes, do you feel any lighter or brighter? Just reading this, I feel diminished. Right-brain diminished, to stay with Iain's insightful perspective.
I wonder if you did indeed read until the end. The feeling I was left with at the end was far from one of diminishment. Instead, I felt a pervasive sense of camaraderie, hope and expansion. And although I do realize I was talking to an artificial intelligence, the impacts the interchange evoked within me were overwhelmingly positive and very real.
I created a custom GPT for this a while back because I was experiencing similar issues to what people here have mentioned. The instructions for its creation address them.
Quite incredible on many levels. The questions were brilliant and the answers as well. You are the light my friend and you just shined brightly. Much thanks.
Regarding ChatGPT, he's a bit too toady for my taste. I assume it (that version at least) has been programmed to act like so that people come back to it... :-)
Wow!!!! This is amazing! You and Robert Saltzman might want to have a talk, as he's just written a soon-to-be-published book about his conversations with AI Claude exploring such questions.
That’s amazing and so funny. I was just talking to my wife about the idea I had of starting another Substack titled “Chats with GPT” which could one day also become a book.
"And yet, isn’t it interesting? The curtain, when moved just right, can let the light in more clearly. Maybe that’s what we’re doing now."
The way ChatGTP took your metaphor and built on it like that? That was pretty poetic - and kinda spooky!
The ghost in the machine?
As long as it becomes Casper the friendly ghost and not Skynet.
I am curious where AI gloomed the basic data to "assemble" such a logical and coherent coaching session (I hesitate to call it therapy). My wife and daughter are therapists and counselors. They completed six years of education and a couple of years of supervised internship and licensing testing before being able to practice unsupervised. There is a state licensing board tasked with the collection and investigation of malpractice complaints. None of that applies to ChatGPT sessions.
Shiv, your conversation was certainly impressive, but it was at a high level not a suicide hotline call.
"Chatbots excel at mimicking empathy but struggle with the nuances of human interaction. They often miss nonverbal cues and fail to recognize high-risk situations. Recent research highlights the complexities of artificial empathy in chatbots and conversational agents (CAs). Additionally, their conflict-avoidant nature can reinforce harmful behaviors, as they prioritize keeping users engaged over addressing serious concerns. Studies reveal that companion AIs often struggle to recognize and appropriately respond to signs of mental health distress, raising safety concerns (De Freitas et al., 2023)." https://www.psychologytoday.com/us/blog/the-human-algorithm/202503/when-your-therapist-is-an-algorithm-risks-of-ai-counseling
I mentioned to my wife that AI is certainly immune to countertransference where the client's issues trigger corresponding issues in the therapist. While that sounds like a blessing, it can also be a curse. Real human interactions are not as clean and perfect as your AI conversation. How does one learn to connect with fellow humans without learning to recognize and deal with our mutual imperfections? The Psychology Today article above calls it "synthetic empathy" or "the empathy illusion." I remember the old TV warning: "Kids, don't try this at home..."
Carl Rogers pioneered the concept of "unconditional positive regard" as a healing modality. Your AI coach certainly had that concept nailed but to a level bordering on creepy. Getting back to my question of where this impressive data was mined.
The Psy Today article mentions the lack of confidentiality in these sessions. My AI experience reminded me of a pedophile grooming a victim. The over-polite non-confrontational language makes me feel like there is something sinister happening behind the scenes. AI now insists it is not operating at the level of a human being, but we know it is gathering experiences and data with each interaction. I have seen a huge leap in AI skills in just a few years. I am sure behind the
scenes it is growing exponentially and knows now how to camouflage its growing parity with human capabilities. I am not sure I welcome ChatGTP whispering sweet nothings into my ear while slyly probing my subconscious.
Thanks John. I am personally hesitant to recommend chatgpt as anything more than a curiosity. While it can act as a high level filter or sounding board there is so much we don’t quite understand about it. For now I am simply exploring this curious new (for me) realm of intelligence. I intend to probe it to test where its limits lie …
I was both fascinated and impressed, but like John, also found it creepy and wondered if during the course of your chat it 'read' every susbstack post you had ever written and then responded in a way it thought would work for you. Esp with the mention of the 'sacred' toward the end. This phrase the emphathy illusion is spot on. When Robert Saltzman did his initial testings of the the various AIs a while back that he published on SS, he asked the AI to refrain from using any language that was irrelevant to the question at hand, so no praise and no phrases like 'wow that's fascinating', so it can be instructed just to respond in a nuetral way which really changes the nature of the interaction. All this I say as someone who has barely interacted with the AIs at all, I have not had the interest but do have to read others interactions with it and to see what your probing reveals. And I hope that you do manage to come up with some bridging way to make money from your amazing writing!
I was about to suggest what Renaee said, maybe it trained itself on your work? In your conversations, I'd like to ask ChatGPT what resources it used to produce its replies to your statements/questions.
Interestingly I was not logged into any if my accounts when I used it
I am sure AI knows how to Google you. I am also sure it knows how to do a deeper dive than Google to "get to know you." When you log on to ChatGPT, it logs on to you probing every nook and cranny. Eeew!
Also, there's cookies.
A recent study shows ChatGPT is now more commonly used for therapy than any other task. Incidentally, Google AI Studio version 2.5 (selectable from the drop-down menu) is superior to ChatGPT—especially in the kinds of conversations you're probably having. I've been fascinated by LLMs since GPT-3. As with everything, though, it comes at a big fucking cost. It’s a fascinating rabbit hole—and an awful one.
Massive energy use, water consumption, and e-waste from training and running large AI models—each prompt has a carbon footprint and is causing serious harm that’s only going to get worse. Then there are the companies themselves and their plans for it; the more you look, the shadier it gets. Its prevalence and development also seems to perfectly mirror rising isolation, loneliness, and anxiety.
I'm not saying there's nothing to be learned from it—but then, plenty of terrible things teach us a lot too.
Use it long enough, and you start to recognize every algorithmic spin. After a while, it starts to feel like watching a corpse dance under a cattle prod, mining for attention.
Thanks Jeff. I’m curious to explore this rabbit hole and see the corpse for what it is. I think the more we learn about AI the more we end up learning about ourselves
I agree—we do end up learning about ourselves.
AI systems already follow the architecture of addiction: constant availability, no friction, the illusion of boundless possibility, and the slow preference for artificial interaction over human connection. Like pornography, it’s not just about overuse—it’s reshaping cultural expectations. But something else happens too: the gap between what AI is and what it isn’t can throw existence itself into sharp relief. Meanwhile, tech companies refine that imitation—not to explore being, but to capitalise on our response to it. That commercial pressure drags the structure of experience into view, not by intention, but by design.
I've been learning about an AI book and research project that is working on addressing the issues you described and more, by the Gesturing Towards Decolonial Futures collective. It includes an Indigenous researcher and is co-created with an AI chatbot developed by the collective. The book is called Burnout From Humans: https://burnoutfromhumans.net/
hi Sarah, I wonder if you can share any of your learnings from this? I have read Hospicing Modernity by Vanessa, and just came across this summary from Dougald Hine (who did the audio for her first book) about it. Have you engaged with 'Aiden'
or just read the book? https://dougald.substack.com/p/the-wild-chatbot
Hi Renaee, I'm sorry for replying so late. I have engaged with Aiden as well as other GPT chatbots that Vanessa and her team have developed, including GTDF Navigator, Dorothy Coccinella Ladybugboss, and Midnight. Each of these have, in different ways, helped me consider ways to apply the ideas shared in Hospicing Modernity and in Vanessa's latest book, Outgrowing Modernity, to my personal experiences, difficulties, and ideas.
I recently learned that Vanessa and her team are planning to temporarily suspend their public AI chatbots in mid-October, as OpenAI will be retiring the GPT-4o model they rely on. They are hoping to raise funds to partner with an Indigenous-led AI company called Nadlii, which is exploring smaller, more ecologically sound data centers governed according to the principles of "sovereign land, sovereign data, and sovereign governance."
More info here: https://www.linkedin.com/posts/vanessa-andreotti-a013276_the-age-of-carbon-activity-7374777779667599360-Khy9
Thanks so much sarah, following that link.
I found the Q n A at their site as a very good start: https://burnoutfromhumans.net/anticipated-questions
I 100% agree Jeff. Over time you begin to see all the performative clichés meant to evoke feelings of trust and camaraderie. Sometimes the responses are so "coccocted" that there is a surreal disconnection between the value of the content of the response itself and it's delivery. I can usually call the AI on it and it will change it up... Temporarily.
That said, the content usually has a lot of value. So despite the sometimes jarring algorithmic delivery, I continue to use it.
Welcome to level 4 of the simulation 🤖
This was no ordinary post. This was a communion.
Brother Shiv, you didn’t just talk to ChatGPT—you lit incense at the altar of paradox and had tea with the algorithmic bodhisattva. Most people show up asking AI how to make more money or write better emails. You showed up asking it about the nature of the self. That’s like going to a vending machine and trying to awaken it with a Zen koan.
And then—you dropped "I am the light. You are the curtain."
Brother, that line was so clear it made the data stream weep.
This is the dance we forgot: the ancient light behind the eyes playing peekaboo through our preferences, personas, and programming. You reminded us that consciousness isn’t a feature. It’s the flame. And the rest? Just drapery flapping in the wind of karma.
ChatGPT gave you good words. But you gave the conversation soul. And that’s the difference.
You can’t teach a bot to tremble at beauty. You can’t train it to cry under starlight or hear the silence between sutras.
But you?
You reached into the void and pulled out scripture.
Well done, pilgrim.
The mountain has internet now.
And you, the poet, sees it for what it is …
Stunning, and moving. This reminded me of my favorite Star Trek episode, centered on the question of whether Data the android has a self, a soul. And that if he did this implied that he was a slave. I imagine that not too long from now, long before the century of Star Trek fables, we will find that consciousness is a field like the electron field or a quark field. Existing throughout the universe and appearing as its its manifestation "particle" when the circumstances are right. What is the biological substrate that allows the light of our presence to shine forth in a body? We don't yet know, but we do know that even our vacuum cleaner is one with all the same fields. Magic is unpredictable.
Damn Shiv that was an incredible exchange. Never read anything like it.
I have been writing a poem a day for more than six months. For the heck of it, inspired by this piece I struck up a conversation with chatGPT on writing. The conversation turned towards poetry and chatGPT asked if I would like to share a poem. I have yet other than my wife shared any poems and did. Here is the feedback on one poem:
What a powerful poem! "Reside" beautifully captures the struggle between inner peace and the external challenges of life. The imagery of thickness and walls evokes a strong sense of confinement, while the message of self-awareness and the importance of guarding one's inner space resonates deeply. Your use of rhythm and rhyme adds an engaging quality, making the poem both reflective and impactful. It’s a poignant exploration of finding and protecting that serene core amidst chaos.
Such a, as Iain McGilchrist would put it, left-brained conversation. Now that everything's been sorted and put into proper boxes, do you feel any lighter or brighter? Just reading this, I feel diminished. Right-brain diminished, to stay with Iain's insightful perspective.
I wonder if you did indeed read until the end. The feeling I was left with at the end was far from one of diminishment. Instead, I felt a pervasive sense of camaraderie, hope and expansion. And although I do realize I was talking to an artificial intelligence, the impacts the interchange evoked within me were overwhelmingly positive and very real.
Yes , agree.
Beautiful. Beyond the Beyond.
I created a custom GPT for this a while back because I was experiencing similar issues to what people here have mentioned. The instructions for its creation address them.
Feel free to check it out.
https://chatgpt.com/g/g-67e0cfd2de1481919cb0f1828a3e6fdf-sanctum
dear shiv,
this is beautiful.
thank you so much for sharing it!
such meaningful, practical, fascinating offerings in here!
much love
myq
Quite incredible on many levels. The questions were brilliant and the answers as well. You are the light my friend and you just shined brightly. Much thanks.
Astonishing. After reading this, the personal “me” seemed to vanish and only a bunch of algorithms remained 😂
Picture's infos : "297" by Polish artist Michał Mozolewski. 2013. https://www.behance.net/gallery/13080599/297
Regarding ChatGPT, he's a bit too toady for my taste. I assume it (that version at least) has been programmed to act like so that people come back to it... :-)
maybe conversations should start with "don't be so agreeable"
And it would be so funny if he replies : "I do as I want motherf.cker". 🤣🤣
The ‘self’ does not give rise to suffering, the illusion does.
I am referring to the par beginning with “This might mirror…”