[Jaro]

Hello everyone,

I would like to share my experiences with AI-assisted Dhamma contemplation with you. I have been experimenting with it a lot lately to get a feel for what is currently possible with this technology.

Meanwhile, OpenAI, Google, and Antropic have not only released very powerful models, but also made several other critical improvements:

• The models can perform deep research. This means that they take a wide range of sources into account with a high degree of accuracy. This results in comprehensive and detailed reports.

• In addition, user-defined instructions can be specified, including the role of the model, its behavior, and its tasks and limitations.

• However, perhaps the most important improvement is that additional context can be provided to the models in the form of text files, PDFs, or similar. This gives the models additional knowledge that is insufficient or completely absent in their training data.

This allows these models to be tailored precisely to your own requirements and achieve significantly better results.

I tested the whole thing with Puredhamma by providing the AI with the most important aspects of Tilakkhana and the five aggregates, as well as discourses by Waharaka Thero in the form of text files as context.

I also instructed the model to use only Pure Dhamma as a source and to view everything through their perspective. My goal was to teach the AI what problems traditional interpretations pose and how Puredhamma fixes them.

I instructed Google Gemini Pro 2.5 to analyze the Vitakkasanthāna Sutta (MN 20), using only Puredhamma’s interpretation of the Dhamma.

I found the results and the interpretation of the five ways of dealing with negative thoughts very interesting. I would be very interested to hear what you think of the answers.

Gemini’s answer:

The First Method: Shifting Perception, Not Just Attention

The sutta begins: “when a bhikkhu is giving attention to some sign, and owing to that sign there arise in him evil unwholesome thoughts… then he should give attention to some other sign connected with what is wholesome”.Conventionally, this is understood as a cognitive behavioral strategy: when a “bad” thought arises, one should distract the mind by focusing on a “good” or “skilled” thought. The simile used is that of a carpenter using a fine peg to knock out a coarse one.   

From the Pure Dhamma perspective, this method is far deeper than mere replacement or distraction. It is an instruction to fundamentally shift one’s perception of the very object or “sign” that triggered the unwholesome thought. The unwholesome thought—connected with desire, hate, or delusion—arises because the mind perceives the object with a nicca saññā, a perception that the object has lasting value and can provide stable happiness. This is the “coarse peg” of ignorance.   

The “other sign connected with what is wholesome” is not an arbitrary positive thought but the direct cultivation of anicca saññā.   

Anicca is not simply impermanence. Its true meaning, as explained in the Girimānanda Sutta (AN 10.60), is one of vexation (aṭṭīyati), shamefulness (harāyati), and disgust (jigucchati). It is the profound realization that nothing in this world can be maintained to one’s satisfaction (”   

yam pi iccam na labhati tam pi dukkham“) and is therefore ultimately worthless and a source of distress.   

Thus, the first method instructs the practitioner to contemplate the anicca nature of the very thing they desire or resent. If lust arises from seeing a beautiful form, one contemplates the vexatious nature of that form—its inevitable decay, its inability to provide lasting satisfaction, its nature as a composite that will fall apart. This is the “fine peg” of wisdom (paññā) that directly strikes at, pushes out, and gets rid of the “coarse peg” of ignorant perception. It is not distraction; it is the application of truth to dispel illusion.

The Second Method: Contemplating the True Danger (Ādīnava) of Saṃsāra

If the first method is unsuccessful, the sutta advises that “he should examine the danger in those thoughts thus: ‘These thoughts are unwholesome, they are reprehensible, they result in suffering'”. The conventional interpretation is to reflect on the immediate negative consequences—psychological distress, social harm, or moral blame. The simile is of a well-dressed young person who would be “horrified, humiliated, and disgusted if the carcass of a snake or a dog or a human being were hung around his or her neck”.   

The Pure Dhamma analysis reveals that the “danger” (ādīnava) to be examined is the profound and terrifying danger of the saṃsāric journey. The true consequence of an unwholesome thought is not just immediate discomfort but the generation of   

kamma that leads to future suffering (dukkha), particularly rebirth in the four lower realms (apāyā), where suffering is unimaginable and escape is nearly impossible. This contemplation connects directly to the second aspect of the   

Tilakkhana, expressed as “dukkhaṁ bhayaṭṭhena“—the understanding that the suffering inherent in this world is something to be truly feared.   

The unwholesome thought, rooted in dasa akusala, is the direct cause that leads to this terrifying future. It is the action that makes one helpless (anatta) in the face of kammic results. The simile of the carcass is thus understood in its full gravity: the unwholesome thought is a “kammic carcass.” Clinging to it is like tying a rotting corpse to oneself, a foul burden that will inevitably drag one down into the foulest and most painful states of existence. This is not a contemplation of social reprehensibility but of the cosmic law of cause and effect, designed to instill a profound sense of fear and urgency regarding one’s own mental actions.   

The Third Method: Disengagement Through Wisdom (Anattā)

Should unwholesome thoughts persist, the third instruction is that “he should try to forget those thoughts and should not give attention to them”. On the surface, this appears to advocate for willful suppression or ignorance, a turning away from the problem. The simile given is of a person with good eyes who, not wanting to see something, either shuts their eyes or looks away.   

However, in the context of Pure Dhamma, this is not an act of suppression but a natural and effortless disengagement born of wisdom. Having already contemplated the anicca (worthlessness) and dukkha (danger) of the thought, the practitioner is now positioned to see its anatta nature. Anatta here means more than “no-self”; it signifies that the thought is essenceless, unfruitful, without refuge, and not worth identifying with. The mind sees the thought as ”   

anattā asārakaṭṭhenāti“—unfruitful and ultimately leading to a state of helplessness.   

When the illusory value of the thought has been penetrated through the first two methods, the mind naturally loses interest. It is no longer compelling. The act of “forgetting” or “not giving attention” is not a forceful act of will but the result of having seen the thought for what it is: an empty, conditioned phenomenon. The simile of looking away is thus an act of wisdom: one turns the “eye of the mind” away because one has already ascertained that there is nothing of substance or value to be found there. This disengagement is the practical application of realizing that the thought is not “mine,” I am not “this,” and this is not my refuge (“etaṁ mama, esohamasmi, eso me attā’ti“).   

The Fourth Method: Observing the Arising of Saṅkhāra (Paṭicca Samuppāda)

If the thought still arises, the fourth method is to “give attention to stilling the thought-formation of those thoughts”. This is often the most obscurely translated method, with interpretations ranging from relaxing physical tension to simplifying the thought into a single word.   

The Pure Dhamma interpretation identifies this as the most direct and advanced application of insight (vipassanā). It involves observing the causal process of the thought’s arising in real-time, which is the process of Paṭicca Samuppāda(Dependent Origination) in action. The practitioner turns their attention from the content of the thought to the process of its formation.

One observes how a sensory contact (phassa) leads to a feeling (vedanā) that is conditioned by a distorted perception (saññā). One sees how this feeling, if clung to, fuels craving (taṇhā), which in turn gives rise to the mental formation (saṅkhāra) that is the unwholesome thought. By seeing this causal chain, one understands the principle of ”   

Hethun patichcha Sambhūthan, Hetu bhangā Nirujjathi“—that which arises from causes will cease with the cessation of those causes.   

This is the cultivation of the wisdom of “arising and attenuation” (Uppāda Vaya). By seeing the “arising” (   

udaya) of the thought as dependent on specific conditions (ignorance, contact, feeling, craving), one understands precisely how to bring about its “attenuation” (vaya) by removing the fuel. One stops feeding the craving, and the thought, deprived of its cause, subsides. This method deconstructs the thought, revealing it not as a solid, monolithic entity, but as a fleeting, conditioned, and impersonal process. This is the pinnacle of wisdom-based practice within the sutta.

The Fifth Method: Forceful Restraint as a Preliminary Tool

Finally, if all else fails, the sutta offers a last resort: “with clenched teeth and the tongue pressing on the palate, he should beat down, constrain, and crush his mind with his mind”. This is universally recognized as a method of forceful suppression.   

The Pure Dhamma framework acknowledges the necessity of such a tool but frames it correctly as a preliminary, not a primary, method. The path to liberation is fundamentally one of dassanena pahātabbā—abandoning defilements through seeing and insight. Forceful suppression, while it may achieve a temporary calm (   

tadaṅga samatha), does not eradicate the underlying anusaya (hidden defilements). Suppressed defilements remain dormant and will resurface when conditions are right.   

Therefore, this fifth method is not the path itself but a tool to clear the way for the path. Its use indicates that the mind is too overcome with agitation for the subtle work of insight (methods 1-4) to be undertaken. It is a means to establish a baseline of mental control, from which the genuine work of wisdom can begin. The need to resort to this method is a sign that the conditions for insight are not yet present, and the immediate task is to stabilize the mind so that wisdom can be applied. It is a pragmatic concession to the reality of an untamed mind, but the ultimate solution always lies in the transformative power of understanding, not in the temporary victory of willpower.

[Lal]

Thank you, Jaro. Very interesting.

• I just scanned through it, and AI seems to have done a decent job of summarizing. 
• We can discuss if anyone has questions.

_______

AI has the benefit of being able to analyze a large amount of text in a short time. 

• However, I have always wondered whether it will be able to “think for itself” and come up with new ideas and discoveries.
• Not relevant to this topic: Will AI be able to make discoveries on its own? For example, consider Newton’s discovery of the laws of gravity or Einstein and others’ discovery of quantum mechanics. My belief is that it will not be. I think it is only a way to summarize large amounts of existing information and provide a summary very quickly. But we will see. I believe this is what they refer to as “general intelligence,” and they (Elon Musk, Sam Altman, Demis Hassabis, etc.) think it can be achieved within a few years.

[Jaro]

Do you think the AI’s interpretation of the sutta is acceptable? In any case, I find the reasoning very clever and plausible, even if I’m not sure whether it might not be a bit far-fetched.

Here is another tabular overview of the interpretations:

Method	Conventional Interpretation (Summary & Keywords)	Pure Dhamma Interpretation (Summary & Core Concepts)
1. Replace the Sign	Replace an unskillful thought with a skillful one (cognitive replacement, distraction). Keywords: replacement, skillful thought, fine peg.	Shift perception by contemplating the anicca nature (worthlessness, vexation) of the object that triggered the thought. The “fine peg” is wisdom (paññā).
2. Examine the Danger	Ponder the immediate, mundane disadvantages and psychological harm of the thought. Keywords: disadvantages, danger, reprehensible.	Contemplate the true saṃsāric danger (ādīnava)—the terrifying dukkha of rebirth in the lower realms (apāyā) as a consequence of the thought.
3. Ignore/Forget	Willfully ignore or suppress the thought; turn attention away. Keywords: forget, not give attention, look away.	A natural disengagement born of wisdom, realizing the thought’s anatta nature (essenceless, unfruitful, not ‘mine’).
4. Still Thought-Formation	Vaguely defined as relaxing tension or simplifying the thought process. Keywords: stilling formation, relaxing fabrication.	Directly observe the causal arising of the thought via Paṭicca Samuppāda, seeing its conditioned nature and thereby removing its fuel (hetu).
5. Forceful Suppression	A last-resort, brute-force method to crush the unwholesome mind with the wholesome mind. Keywords: crush, beat down, restrain.	A preliminary tool to establish temporary calm (tadaṅga samatha) when the mind is too agitated for insight. It is not the path of wisdom (dassanena pahātabbā).

I found the fact that method 5 is an absolute emergency solution particularly interesting.

As for your question as to whether AI will ever be able to derive new insights from existing knowledge, I can only speculate.

In any case, the current models are not in a position to do this, as they are essentially trapped within their universe, i.e. the totality of all training data. This makes it impossible for them to think outside the box and be creative and innovative.

It is difficult to predict what the future holds. I find the rapid progress in the field of artificial intelligence both impressive and frightening. 

I believe that models that are able to create completely new theories are possible in principle. Every scientist has developed their own theory based on other theories and research findings.

One example is Schrödinger’s theory, in particular his famous wave mechanics and the associated Schrödinger equation. It was inspired by Louis de Broglie’s idea of matter waves and Hamilton’s classical mechanics.

If you train a model with certain initial theories (input) on a target theory (output), it could in principle learn how to gain new insights from existing knowledge.

At least this is a naive approach that could perhaps be tried out. It would merely be a mapping from the source of inspiration to the resulting theory.

But if we are honest, there are already countless theories in physics. However, most of them fail to be validated against measurement data from the real world.

Despite all the hype, it should not be forgotten that current models are statistical in nature and language modelling is not intelligence. 

[Lal]

Yes. I think you summarized the current capabilities of AI well in your following statement: “In any case, the current models are not in a position to do this, as they are essentially trapped within their universe, i.e., the totality of all training data. This makes it impossible for them to think outside the box and be creative and innovative.”

• As I also mentioned, AI can only summarize, but it does it well.

___________

You wrote: “I believe that models that are able to create completely new theories are possible in principle. Every scientist has developed their own theory based on other theories and research findings.”

• Yes. However, when examining the “chain of progression” (where one scientist builds upon the previous accomplishment of another), all of them contributed an original idea to help the process move forward. That progression eventually leads to a significant breakthrough by a scientist who is fortunate to have the previous accomplishments of others to build upon. But that last step may require an especially bright mind.
• What I tried to say is that AI (as of now), cannot come up with such an original idea (breakthrough concept) anywhere in that “chain pf progression.” 

P.S. Regarding your first question: “Do you think the AI’s interpretation of the sutta is acceptable?”

• Yes. AI seems to have summarized both versions reasonably. Of course, it does not compare the two to make a judgment on which is better.
• I wonder whether it will do so in a more complex concept, like Tilakkhana.

[Dhammañāṇa Bhikkhu]

Good householder,

It’s not possible to gain Dhamma if not given and received in proper ways. While machines may serve desires and defilements well, their lack of goodness, and for the most even the opposite, since it’s clear that sources are not even legal rightly given, no need to tell that such could never happen in line of Dhamma.

 

It’s a waste to try to gain relation with the Gems ouwardly their domain.

(Samana Johann)

[Author]

Posts