TAROTVIBES & AI

How the AI behind the readings works, its limits and responsibilities

DOES TAROTVIBES REALLY USE AI?

Yes. Every reading you see on screen is generated in real time by a generative AI model that produces original text from your inputs. There are no pre-written interpretations pulled from a database: every paragraph you read is composed on the spot, based on the randomly drawn cards, your free-text question and your zodiac sign. Two different users asking the same question would get different answers, and you yourself, repeating the same question, will get different interpretations — because the cards change and because the model introduces narrative variation.

HOW IT WORKS IN PRACTICE

The technical flow has four steps:

• Random card drawing. When you press "Draw the cards", the server uses a cryptographic PRNG (not a simple random()) to pick 1, 3 or 5 cards from the 22 Major Arcana without repetition, also deciding whether each is upright or reversed. This means the draw is neither tamper-proof nor predictable.
• Prompt construction. The system combines the drawn cards (with their traditional symbolism), your free-text question, the chosen category (love, work, future, decisions) and your zodiac sign to build a structured prompt that guides the AI model.
• Reading generation. The AI model receives the prompt and produces an original narrative reading, in your language, that integrates the cards' symbolism with the question's context. The model is instructed to keep a reflective tone and avoid absolute claims about the future.
• Saving to history. If you are logged in, the reading is saved to your private history. You can re-read it, add reflections, and delete it whenever you want.

WHAT THE AI DOES **NOT** DO

It is important to be clear about what TarotVibes' AI does not do, because the boundary is often misunderstood:

• It does not predict the future. No statistical model or LLM has access to information about the future. The AI reorganises traditional symbols around your question, but it knows no more about tomorrow than you do.
• It does not read your mind. The AI only knows what you tell it: the question, the category and the zodiac sign. It does not analyse your browser, profile your behaviour or aggregate data from third parties.
• It does not replace a therapist. If you are looking for psychological support, AI is not the answer. A human therapist has empathy, context and training no model can replicate.
• It does not give medical, legal or financial advice. Interpretations are symbolic prompts for reflection, not professional recommendations. For important decisions, consult a qualified professional.
• It does not manipulate results. Cards are drawn with the cryptographic PRNG before the prompt is generated. There is no "bias toward positive readings" or "toward dramatic readings": the model works with whatever chance decided.

WHY IT WORKS AS A REFLECTION TOOL ANYWAY

TarotVibes' usefulness does not lie in "knowing the future" — it doesn't — but in offering an external symbolic frame onto which to project your question. It is a well-studied psychological mechanism: when we look at a rich metaphor (tarot cards are, by definition, rich — built and refined for centuries), we tend to identify aspects of our present we hadn't noticed. It works like a narrative Rorschach: the "truth" of the reading is not in the card or in the AI, it is in the reaction it triggers in you. And it is a useful tool, as long as you know that is what it is.

LIMITS AND ETHICS

We chose to expose TarotVibes' AI with explicit disclaimers because in a borderline space like divination, ambiguity is harmful. Vulnerable people — those going through grief, anxiety, depression, hard decisions — can develop emotional dependence on a forecasting tool perceived as "intelligent". That is why:

• Every reading restates the entertainment framing.
• We do not gamify "winning" positive readings: badges reward consistency and curiosity, not outcomes.
• We do not sell upsells of "more accurate premium readings": all readings use the same model.
• We do not promise accuracy or show fictional "reliability scores".
• Disclaimers in the Terms of Service are explicit, not buried in legalese.

PRIVACY AND DATA USED BY THE AI

The AI only receives the data it needs to generate the reading: the drawn cards, your free-text question, your zodiac sign and the chosen category. It does not receive your email, your IP, your previous history. We do not train models on the content of your questions. Saved readings stay in your private history, readable only by you (no one else, not even us, reads them except for explicit technical issues). You can delete them whenever you want.

DEEPER DIVES