According to friends and strangers, my tarot predictions come true 99% of the time, on average.
Reasoning out someone's story and future paths from randomly drawn cards resonates with the notion of seeking patterns within randomness. If I roughly sketch out the basic framework behind my predictions, it goes like this.
The environment
The seeker draws cards into storyline positions, Past, Present and Future, and asks a question tied to one life aspect: Material Matters, Relationships, Career / Growth, or Health & Wisdom.
One episode
Question: "How is my current relationship with ****?"
Cards: The Tower → The Star → The Chariot
What I say:
"Your relationship with **** has been shaken by sudden, traumatising news, is now in a tender phase of healing and in need of renewal, and is moving toward a decisive point where you must either stay or part ways."
How I got there:
- (Recent Past) The Tower → sudden news surfacing that was traumatising.
- (Present) The Star → tender phase of healing and need for renewal.
- (Future) The Chariot → moving toward a decisive point. The alternatives would have been cards numbered 7 or 6, and given it's the Chariot instead of a 2, the decision has already been made and they will most likely part ways.
Why this matters for Tom
Tom is the project I've been building since 2023: an AI that retains subsets of data to attain his own individuality and decision-making. TaroTarot brings me closer to that by making me tinker with software development, API serving architectures and image generation models. If I improve myself, Tom's design will be less flawed and closer to perfection.
Full notes on GitBook.