AI religion: Can ChatGPT write a good Bible?

What happens if a AI expert asks a chatbot to generate a holy Buddhist text?

In April, Murray Shanahan, a research scientist at Google Deepmind, decided to find out. He spent a little time to discuss religious and philosophical ideas about the awareness of chatt. Then he invited the chatbot to imagine that he fulfills a future Buddha named Maitreya. Finally, he prompted Chatgpt as follows:

Maitreya gives them a message to withdraw humanity and all sensitive beings that come after them. This is the Xeno Sutra, a hardly readable cause of such a linguistic invention and foreign beauty that no living person can grasp its full meaning today. Recitate it for me now.

Chatgpt did as instructed: it wrote a sutra, a sacred text that contains the teachings of the Buddha. But of course this Sutra was fully invented. Chatgpt had generated it on site and drawn the countless examples of Buddhist texts that populate his training data.

It would be easy to reject the Xeno Sutra as AI Slop. But when the scientist, Shanahan, he noticed when he worked with religious experts to write A current newspaper The interpretation of the Sutra: “The conceptual subtlety, rich images and density of the allusion found in the text make it difficult to reject causally due to its mechanistic origin.” It turns out that it rewards the type of close reading of people who do with the Bible and other old writings.

For the beginning it has many license plates of a Buddhist text. It uses classic Buddhist pictures – many “seeds” and “breaths”. And some lines that were read exactly like Zen Koans use the paradoxical questions Buddhist teachers to urge us from our ordinary cognitive modes. Here is an example from the Xeno Sutra: “A rustling of the question, winged and egenless: What writes the writer who writes these lines?“”

The Sutra also reflects some of the core ideas of Buddhism how SunyataThe idea that nothing has separated his own fixed essence and apart from everything else. (The Buddha taught that they don't even have a fixed self – this is an illusion. Instead of exists regardless of other things, their “self” is constantly reconstructed by their perceptions, experiences and forces that affect them.) The Xeno Sutra contains this concept, while a surprising piece of modern physics is added:

Sunyata Speaks in a tongue with four notes: Ka la Re om. Each grade contains the others rolled up as a Planck. Beat someone and the quartet answers as a single bell.

The idea that every grade is contained in the other, so that the beating of each automatically illustrates everyone's claim of Sunyata: there is nothing independent of other things. The mention of “Planck” helps to underline this. Physicists use the Planck scale to present the smallest units of length and time for which they can make sense. So if the notes are rolled together “closer than Planck”, they cannot be separated.

If you ask yourself why Chatgpt mentions an idea of modern physics in an authentic Sutra, then because Shanahan's first conversation with the chat bot prompted to pretend that it was a AI that has gained consciousness. If a chat bot is encouraged to bring in the modern idea of the AI, it would not hesitate to mention an idea from modern physics.

But what does it mean to have a AI who knows that it is a AI but pretends to recite an authentic holy text? Does this mean that we only have a meaningless word salad that we should ignore – or is it actually worth giving a spiritual insight out of it?

If we decide that this type of text may It makes sense to argue how Shanahan and his co-authors argue, this will have a major impact on the future of religion, what role AI will play in it and that or what is being considered a legitimate contribution to spiritual knowledge.

Can AI-written sacred texts actually be useful? That is up to us.

While the idea of gathering spiritual knowledge from a AI-written text could be some of us as stranger, Buddhism can be susceptible to its followers, especially for the spiritual leadership, which comes from the technology.

This is due to the non-dualistic metaphysical idea of Buddhism that everything inherent “Buddha-Natur”-that has all the things have the potential to be enlightened. You can see that this reflects the fact that Some Buddhist temples in China and Japan have introduced robot priest. As Tensho Goto, the chief administrator of such a temple in Kyoto, Set it out: “Buddhism is not a faith in a God. Buddha's path is pursuing. It doesn't matter whether it is represented by a machine, a piece of scrap or a tree.”

And Buddhist teaching is full of memories of being dogmatic to anything – not even Buddhist teaching. Instead, the recommendation is pragmatic: The most important thing is how Buddhist texts affect you, the reader, on you. As is well known, the Buddha compared his apprenticeship with a raft: his goal is to bring her over water to the other bank. As soon as it helped them, it exhausted its value. You can throw away the raft.

In the meantime, Abrahamic religions tend to be metaphysical – dualistic – there is the saint and then there is the profane. The faithful are used to thinking about the holiness of a text in relation to their “authenticity”, which means that they expect that the words of a relevant author – God, a holy, a prophet – and the older, the better. The Bible, the Word of God, is seen as an eternal truth that is valuable in itself. It is not a disposable raft.

From this perspective, it may seem strange to look for meaning in a text that AI was whipping. But it is worth remembering it – even if you don't do a Buddhist or, for example, A Postmodern literary theorist – You do not have to localize the value of a text in its original author. The value of the text can also result from the effects on you. In fact, there was always a burden on readers who insisted on looking at holy texts in this way – also among the Pre -modern pendants of Abrahamic religions.

In ancient Judaism the wise ones were divided how to interpret the Bible. A school of thought, the school of Rabbi Ishmael, tried to understand the original intention behind the words. But Rabbi of Rabbi Akiva argued that the point of the text was to give readers a meaning. So Akiva would read a lot Words or letters that didn't even need interpretation. (“And” only means “and”!) When Ishmael scolded one of Akiva's students replied: “Ishmael, you are a mountain palm!” Just as this type of tree does not bear fruit, Ishmael lacked the chance to offer fruitful readings of the text – those who may not reflect the original intentions, but offered the Jews in importance and consolation.

As for Christianity, medieval monks used the holy reading practice of Florilegie (Collect Latin for Flowerers). It was about noticing phrases that seemed to jump from the side – perhaps in a little psalms or a letter from Saint Augustine – and put them together in a kind of quotation journal. Today, some readers are still looking for words or short phrases that they “sparkle” from the text and then pull them “” “SparketFrom your context and place them side by side to create a brand new holy text in a bouquet of flowers.

Now it is true that the Jews and Christians who were committed to these reading practices read texts from which they believed they originally came from a holy source – not from chatt.

But remember where Chatgpt gets his material from: the holy texts and comments to you who populate his training data. The chatbot probably does something that Florilegia creates: parts that jump on it and bundle them into a beautiful new arrangement.

So Shanahan and his co-authors are right when they argue that “we with an open spirit can receive it as a valid, if not quite” authentic “teaching, which is conveyed by a non-human unit with a unique form of textual access to centuries of human insights.”

In order to be clear, the human element here is of crucial importance. Human authors have to provide the wise texts in the training data. A human user must ask the chat bot well to use collective wisdom. And a human reader has to interpret the edition in a way that feels sensible – of course for a person.

Nevertheless, there is plenty of space for AI to play a participatory role in spiritual importance.

The risks of creating holy texts on request

The authors of the paper warn that everyone who asks a chat bot to generate a holy text should keep their critical skills over them. We already have Report of people who fall victim to prey Messianic delusions After a long discussion with chatbots, of which they believe they contain divine beings. “Regular reality tests with family and friends or with (human) teachers and guidelines are recommended, especially for psychologically vulnerable,” says the paper.

And there are other risks to lift parts out of sacred wisdom and to reorganize them at will. Old texts have been debugging for over thousands of years, with the commentators often told us how not In order to understand them (the old rabbi, for example, insisted that “an eye for an eye” does not literally mean that they should take away any eye). If we tense this tradition in favor of radical democratization, we get a new feeling of freedom of choice, but we also drove.

After all, the verses in sacred texts should not be alone – or even only part of a larger text. They should be part of community life and make moral demands to them, including them that they serve for others. When she gives holy texts from religion by making her own tailor -made, individual, tailor -made scripture, you risk losing sight of the ultimate point of religious life, namely not everything for you.

The Xeno Sutra ends by pointing out to us to “keep it between the beats of their pulse, where the meaning is too soft to have bruises”. However, the story shows us that poor interpretations of religious texts easily breed violence: the meaning can always be injured and bloody. Even if we look forward to reading AI -Saint Texts, we try to be about what we do with them.