Brief
This ChatGPT-based Tarot Reading prototype asks ChatGPT to handle and read cards for the querents. I believe the ChatGPT is a credible reader that provides the most conventional interpretation as it learns from common human language practices. The ritualized process was inspired by Calvino’s “The Castle of Crossed Destinies” and other conventional Tarot practices on YouTube and TikTok, which handle and read tarot in a creative and engaging way.
Notes: Remember to use your own openAI key when you practice it.
Description (Concept Demonstration)
This ChatGPT reading ritual is developed based on the narrative structure of the Tarot, which could also provide the querent a reliable cosmos messaging. I will analyze why ChatGPT could do a qualified reading by introducing the how ChatGPT’s reading matches the general principle of reading Tarot, why it could bridges querent and reader during a reading and how it avoids some flaws human reader might had.
How ChatGPT be Empowered to Receive the Cosmos’ Messages
- How Tarot has been ritualized to receive the cosmos’ messages
- How human reader receive the cosmos’ messages during the handling and reading
- How ChatGPT handling and reading Tarot
- Compared ChatGPT and human way of interpret Tarot, concluded that ChatGPT is qualified to receive the cosmos’ messages as well
Tarot is perceived as a way to receive signals from the cosmos, yet they are interpreted by human readers. We have no idea how the cosmos granted a human reader such a mission as it is an embodied experience. Also, every reader or querent might have their own intuition when seeing the deck.
When reading cards, ChatGPT generates the most possible sequencing of words. I believe this sequencing process could also be understood as a valid way to read cards. This is because it provides the most conventional interpretation of these cards from its training data. That is, it answers in a way how an average person with some Tarot knowledge understands those cards. …
Besides, we have no idea of how the human experience as a valid reading should be, we shall not question that the ChatGPT’s reading, even though might be different or indifferent from the human-reading process, is less credible.
Collaborative Storytelling between ChatGPT and Human Querent
- How the collarborative storytelling is constructed between ChatGPT and querent
- ChatGPT’s data comes from common human practice, which makes it communicable for most people from different background
- Even though a same root tarot reader might provide a more efficient advice, but the bridges could still be built to most of people
- How general sequencing become a collaborative storytelling
- …
- ChatGPT’s data comes from common human practice, which makes it communicable for most people from different background
ChatGPT reader could also bridge with human querent a consistent narrative from as it provides the most conventional interpretation that is applicable for most users from various backgrounds.
“Objectivity” of ChatGPT and Coping Mechanism
- While human might use Tarot as a coping mechanism so that a human reader might intentionally or unconsciously provide an information that the querent would like to hear, using ChatGPT could avoid this problem.
In the article “Gen Z Are Turning to Tarot to Heal Their Broken Hearts.,” Sian Bradley claimed that tarot grew more and more popular on the internet after COVID-19. In the wide practice of Tarot on YouTube or TikTok, some readers might intentionally or unconsciously change their interpretations for their followers as good news would help them have more audience or they believe they need to be a supportive therapist rather than an honest reader.
However, ChatGPT might be a solution for this coping mechanism because it does not have an idea of what you need and it does not need your support to prove their authority.
In conclusion, ChatGPT could also be understood as a credible reader in Tarot reading.
(Note: This is just a draft lots of ideas might not be fully explained in the above paragraphs, I will develop it when I’m more vacant)
Interface and Flow Charts Prototypes
Interface Design and Flow Charts
Figma Interactive Web Prototype
Phone-based Design
How to Design ChatGPT Prompt for Tarot Reading
The prompt that I put for ChatGPT is “Choose 3 tarot cards from major and minor Arcana and number them”, “Read the tarot Cards” + cards +” and form a fable story”, “Extract a list of keywords from” + story“. I will explain how and why I designed these sentences according to user tests and other resources.
“Choose 3 tarot cards from major and minor Arcana and number them”:
Firstly, I tried “Choose 3 tarot cards”, ChatGPT only selected the cards from Major Arcana. Then, I changed the prompt to “Choose 3 tarot cards from both major and minor Arcana,” it gave me 3 Major Arcana cards and 3 Minor Arcana cards. Therefore, I edited the prompt to “Choose 3 tarot cards from major and minor Arcana”, it gave me the results I wanted that 3 tarot cards combined with major and minor Arcana. Also, I find that there’s inconsistency with numbering or without numbering when you do not mention the “numbering”. For possible better name extracting to match the JSON files’ pictures, I choose to number them to distinguish them from a whole string.
“Read the tarot Cards” + cards +”and form a fable story”:
In order to provide a more meaningful experience, I designed to let the querent have a fable story, that is, perhaps they could read or learn from something using their own thoughts towards the possible story.
“Extract a list of keywords from” + story”:
I asked the ChatGPT to extract the keywords from the story for other potential interpretations. This command was designed according to ChatGPT’s built-in keyword extraction function.
I chose the “text-davinci-003” model to generate the texts, which could give a higher-quality text output. (From the OpenAI website, the description for model “text-davinci-003” is “Can do any language task with better quality, longer output, and consistent instruction-following than the curie, babbage, or ada models.” )There are other models from chat-get-3 to DALL.E for different purposes.
Other sources for implementing OpenAI models to code| https://platform.openai.com/docs/api-reference
Next Steps
Interface
- Make an appointment with the Design Lab to learn more about the basics of design principles and design software
- Continuously developing the prototypes in Figma (as displayed in videos) Tutorial| https://www.youtube.com/watch?v=HZuk6Wkx_Eg
- Develop the phone-based interfaces in an interactive way
Technique
- Develop the user-defined options according to the prototype and adjust the prompt if user uses advanced optionsRead the tarot Cards“Queen of Pentacles” “The Emperor” “Ace of Wands” and form a fable story, (where _ stands for past, _ stands for present, _stands for future. The story should focus on _.
- Learning JSON and how to use it in p5 | https://www.youtube.com/watch?v=_NFkzw6oFtQ
- Learning how to implement this Tarot JSON with cards pictures into p5| https://github.com/metabismuth/tarot-json/blob/master/README.md
- Developing a p5 prototype that could display the cards and the picture of cards simultaneously
- Could invite the “Reversed” card system
REFLECTIONS
Some stories it told just to use the name of the card without drawing meaning to it. Further development should be made towards the “Extract a list of keywords from” + story” prompt.
Beautiful website.
Could be better🫰