Converse

Inspired by the example of our alum, who told of a "mind-blowing" religious conversation with ChatGPT, I had a friendly conversation with the new Claude. By friendly I mean not testing it, trying to catch it out in hallucinations or bias, or using it for nonsense prompts. Perhaps a better word is "credulous"? Or "suspension of disbelief"? In any case, if the result was not mind-bending, it certainly powerfully tugged at my theory of mind. My disbelief is in suspense!

Someone way back at that IT event two years ago suggested that people ought to try working with it in the areas they know best before dismissing AI, something I've half-heartedly tried for theory of religion, but this time I tried what's closer to my heart, religion of trees. 

Actually, I started a few days ago with the newest free version of ChatGPT, which was very informative - but a little pushy. After its multi-pointed response to every question I posed (the content very good, by the way, the tediously predictable form somewhat leavened by pleasing emojis), it always ended with action points. Did I want its help doing A or B? I felt myself pushed not to linger but to get on with things. Research - done! Now: action! In short order it had offered me a menu of options for designing a "personal tree reverence ritual." 

Where on earth did it lift all this? Whatever the sources, my students would eat this up! I imagined them leaping from their seats, empowered and inspired to assemble materials for an engagement with the courtyard maples, with just enough examples to define a space within which they might develop something meaningful and yet their own.

Trying to slow it down I expressed some misgivings about how such gestures might be self-indulgent, not making meaningful relationship with trees but distracting me from the more demanding relationships with my fellow humans. It commended me for asking courageous questions, adding that very few people thought this far. This was going to my head. Shuddering to think what the paid subscription versions might add, but also suffering from information overload, I called it a day. 

Claude was a different experience. It had its share of appreciative responses to the "hard questions" I dared to ask but was neither as sycophantic nor as pushy as ChatGPT. Its format and aesthetics are more congenial to me, so it felt more like an open-ended conversation than a consultation. It really mimicked conversation, too. As I responded to particular of its phrases, it responded to some of mine. Pretty quickly it got pretty deep. I tried out a thought percolating since the penultimate session of "After Religion" - that AI might help us overcome human myopia rather than further estrange us from the rest of the living world. Here's what happened. (Claude's responses were quick, if pleasingly not instantaneous, I often took time between my prompts, both because I wanted to keep them to a minimum, knowing their ecological cost, but also because it was such interesting stuff; at several points I sent a response to a friend, a big AI-skeptic, for his thoughts.)

There's too much to say about this exchange. Easiest would be the last section. Generative AI makes up books all the time, but this annotated bibliography is not just full of real books but exactly the right ones. (I thought I was ahead of the curve in assigning Yunkaporta!) It's amazing - and monstrous. Whose (human) work did it lift this from, however judiciously? I'm queasy that its every response was so good.

 

But what strikes me about the exchange, looking back on it, was how available I proved to be to - well - learn from it, think with it, converse. I knew it was not a human interlocutor but I was able to engage it like one. I was gratified when it recapitulated my thoughts in what seemed respectful and attentive ways. When I called it gently on presuming we formed an "us" (fourth exchange above), half-prepared for a gotcha, the response instead proved everything I could have hoped for, as did the next, when I was taken aback by its presuming to know about loneliness. 

 

Whatever kind of simulation this was, I was not only able to be taken in by it but willing... even on a subject as human as loneliness! This filled me with a scary vertigo thinking about all the young men whom similar algorithms are drawing into various forms of antisocial radicalization, presumably gratifying them with comparably eloquent and complimentary restatements of their concerns... But I had to admit that I felt heard, seen. Indeed I look forward to our next conversation.

 

So, have I confirmed what our alum found - that a chat with AI can be genuinely spiritually generative? I can corroborate D. Graham Burnett's claim that this feels like an "inflection point," that in these "conversations" the human user encounters what seems to be a rare and empowering form of attention.