AshvinP wrote: ↑Wed Sep 16, 2026 8:33 pm Right, and it can be a bit deflating to one's motivation to dialogue when you know the person on the other side is simply feeding your posts into the LLM and spitting out responses within a matter of minutes, especially in this domain where the ideas require much more careful, participatory contemplation. There is clearly a gradient of augmentation; for example, I am currently using "Grammarly" software to suggest better punctuation, wording, and phrasing. But the trend today is toward full-blown LLM construction. Just today, I commented on an Anthroposophy Facebook post about AI issues that was (admittedly) constructed with AI assistance:
"It is worrisome that even people cautioning about the risks of AI, or the need for ‘spiritual development’, so extensively and casually use LLMs to fashion their posts, and that the trend is only becoming more and more prevalent. This is the primary threat that generally isn’t included in lists of existential AI risks (which focus on physical or psychological, rather than spiritual, harm). Most people simply don’t realize what they are inwardly losing when they fail to sustain their attention through the unfolding of their creative inner process, integrating feedback on their inner gestures as they navigate and articulate the landscape of meaning. I addressed this previously with a Chess metaphor:
"Modern chess software often allows players to take back their moves if the opponent agrees. Many people automatically convince themselves that it is ‘compassionate’ to allow such takebacks. That is somewhat understandable in the case of misclicks, which are usually easy to spot, but what about when the player fails to remain focused and makes an impulsive or ill-thought-out move, only spotting the error after the fact? Allowing a takeback is only ‘helping’ this player if the goal is to win the game, regardless of how it is won. If the goal is to enhance the player’s experience of their intuitive process, on the other hand, then it is optimal for them to live with the full consequences of their cognitive errors, i.e., the suboptimal board configuration that results, so they can properly observe and integrate the feedback. The player is then also given the opportunity to adapt to the new circumstances by discovering creative ways to rebalance their position. The implicit offense here is that, if we are not consistently exposed to the deviations and detours our intuitive process takes from its desired state, in full consciousness, our sensitivity and orientation to the intimate (subtle) constraints on that process can never expand."
Extensive LLM use can be directly compared to this takeback mechanism in Chess, and it is also a vicious self-reinforcing cycle. The more we rely on LLMs to construct our posts, the less sensitive we become to what we are losing by failing to 'manually' organize and articulate our thoughts, observe that process, and creatively navigate the resulting suboptimal configurations. This is an absolutely vital process for modern spiritual development, but it is also a very subtle process and easy to lose sight of. The wider materialistic culture was always expected to keep it in the blind spot, but when it also manifests extensively in Anthroposophically informed spiritual culture, it should prompt us to pause and take stock of what we are doing.
I don't intend to suggest any rigid rules for AI use. There may be times when we need to write something for work and simply don't have the time to do it manually, without a lot of assistance. The key thing is to remain completely self-conscious of what we are doing and not foreclose on opportunities to remain intimately involved in the creative inner process, especially when we are conveying deeper spiritual intuitions. That applies whether we are communicating these intuitions only to ourselves or to others. Extensive reliance on AI for daily writing can easily creep into this spiritual domain and become an unquestioned crutch before we know it. The worst part is that we won't actually realize what we are losing by that overreliance, since we have become so insensitive to the inner process and its higher potential."
Ok, yes, curiously enough we wrote about the same thing. To have philosophical discussions with an LLM in disguise is not only deflating but sad. The other persons soul becomes covered with opaque patches. It's like discussing with a spotted soul, or with a phantom soul.
True, there is a gradient. It's a personal opinion, but for my part I am wary of suggestions for better words or phrasing. I don't ignore those suggestions in work emails, but here I will not do it, despite the issues with my words and phrasing. What does "better" mean in "better phrasing"? Are you pursuing better phrasing here with the intention of being better understood by others? How would AI phrasing be better than your original?
Anyway, yes, what I was complaining about is rather the full-blown type, including when it's only for a paragraph, or for writing a conclusion. In a strange way this trend is helpful because, as a contextual sensitivity for detecting AI language is developed, the occult value of language may become more approachable.