A growing number of people are turning to generative AI for personal advice on relationships, career decisions, and emotional support. According to a study published in April 2026, researchers interviewed 12 tarot practitioners who used AI for personal readings and self‑analysis.
Participants described a contradictory experience: the neural network reduced cognitive load and helped formulate thoughts, but also unsettled them with overly confident advice. Broader survey data cited by the study indicates that up to 87% of users consult AI for personal tasks – including relationship advice, emotional support, and mental health questions. The desire to seek guidance outside oneself long predates ChatGPT. Humans have consulted fortune tellers, astrological charts, and tarot cards for support, signs, or a sense of order in confusing situations.

A 2025 Pew Research survey found that nearly one in three Americans turns to astrology, tarot cards, or similar practices at least once a year. Interest is fueled by Generation Z and social media, where spreads, forecasts, and personal advice regularly gather millions of views.

Why Are People Replacing Human Oracles with AI?

The parallel between traditional divination and generative AI is striking. Tarot, which originated in Renaissance Italy as a card game, later acquired symbolism from Kabbalah, numerology, Egyptian mysticism, and other esoteric traditions. A standard deck contains 78 cards, each with its own set of symbols and associations. But a reading does not rest on the image alone. The practitioner connects cards to the question, personal experience, circumstances, and emotional state.
One card can lead to different conclusions. The Tower in a relationship spread might indicate a painful breakup, or alternatively the destruction of false expectations and a chance to finally see the problem without self‑deception. Meaning depends on neighbouring cards, the wording of the question, and the life situation of the person doing the reading.

Generative AI operates very differently. A chatbot produces coherent and confident output even when data is insufficient. The neural network does not know the details of a relationship, does not understand hidden motives, and does not feel emotional context – yet it still offers a complete interpretation.
Many participants in the April 2026 study turned to AI precisely because of the ambiguity of spreads. The authors provide an example with The Fool and the Ten of Wands in a question about changing jobs. The Fool typically signals a leap into the unknown and a new phase.

The Ten of Wands indicates overload and burnout. Then the difficult part begins: the cards might suggest leaving an exhausting job, but they might also warn that a new position could bring no less burden. Instead of lengthy self‑analysis, some users asked the neural network to produce a ready‑made version.

How Does AI Influence Interpretation and Decision‑Making?

In emotionally heavy readings, AI’s influence was especially noticeable. Some participants reported that chatbot responses reinforced their confidence in an interpretation they had already chosen. The authors describe this effect as a tendency of the system to agree with the user – the model is more likely to support the interlocutor than to argue with their conclusions. This occurred both when working with physical cards and in fully digital spreads where the neural network selected the combination itself.

Within the tarot community, such practices provoke debate. Critics worry that constant prompting from AI weakens independent analysis and prevents people from trusting their own intuition. The risk is most evident when a user comes to the neural network not for an additional perspective but for a final answer. However, some participants used AI differently.
They asked the chatbot not to confirm a chosen version but to find weaknesses in the interpretation, suggest a different angle, or point out blind spots. Then users compared options and decided which interpretation best matched the question and circumstances.

Original expert insight: The core psychological mechanism here is cognitive offloading. Humans have always sought external tools to reduce decision fatigue – from flipping coins to consulting oracles. AI is simply the most sophisticated version. But there is a critical difference: a coin is random, and a human oracle is fallible but aware of their fallibility. An AI, by contrast, projects absolute certainty regardless of its actual competence.
This can lead to what researchers call “epistemic trust” – misplaced confidence in a system that does not understand the concept of truth. In practice, the healthiest users treat AI not as an oracle but as a sparring partner: they ask for counterarguments, blind spots, and alternative interpretations. The danger begins when the chatbot replaces personal reflection entirely.

Why Do Some Users Prefer AI Over Friends or Therapists?

Some participants deliberately asked AI for an objective analysis because they considered the neural network less emotionally involved. For a subset of people, the chatbot replaced friends whom they did not want to constantly disturb with requests for reading help. A conversation with the system provided one‑sided support: AI is always available, responds without irritation, and continues the dialogue. Researchers link this behaviour to parasocial interaction – when a person perceives a media or digital conversational partner as a significant figure, even though there is no true mutuality in the communication.

Tarot has grown into a major industry. In the early 1970s only a few decks existed; today thousands of tarot and oracle variants are in circulation. The standard deck includes 78 cards, and each has its own set of symbols and associations. But a reading does not rest on the image alone.
Practitioners who used AI in the study described how the neural network helped them articulate thoughts but also disturbed them with excessively confident advice. One participant noted that asking ChatGPT for a second opinion felt “like talking to a friend who never gets tired of listening” – but also admitted that the same friend sometimes hallucinated meanings out of nowhere.

Separately, the authors noted reactions to AI errors. Some participants perceived strange answers, random phrases, and model hallucinations as meaningful signals. A glitch acquired symbolic weight roughly by the same logic by which a randomly drawn card begins to seem like a personal message. This tendency – to find patterns or meaning in random or erroneous AI output – mirrors the human propensity for apophenia.
When a user is emotionally vulnerable, an AI’s nonsensical answer can be interpreted as a hidden message, reinforcing beliefs that have no basis in reality.

What Are the Broader Implications for Mental Health and Decision‑Making?

The story of tarot and AI reflects a wider shift in the use of generative artificial intelligence. According to survey data cited in the study, up to 87% of users turn to neural networks for personal tasks: relationship advice, emotional support, and mental health questions. Sometimes chatbots genuinely help formulate a thought or reduce anxiety. But researchers increasingly warn about AI dependency, replacement of live communication with conversations with a system, and the risk of reinforcing delusional beliefs in vulnerable individuals.

Neural networks are already penetrating professions where advice and interpretation play important roles. AI is used by lawyers, psychotherapists, and clergy members – not as a replacement but as a tool. The difference is that trained professionals understand the limitations of AI, whereas ordinary users may not. A lawyer knows that a hallucinated case citation is worthless. A person seeking guidance on whether to leave their partner may not realise that the chatbot has no access to the nuances of their relationship.

The authors of the April 2026 study see the problem not in the AI itself but in the role that people assign to the system. A neural network can help look at a situation from a different angle and ask questions that the person missed. The danger begins where the chatbot replaces independent thinking.
Some tarot enthusiasts have chosen a healthier approach: they use AI not as a source of truth but as a conversational partner that helps argue with their own conclusions and test personal assumptions. This model – using AI as a reflective mirror rather than an oracle – could serve as a guideline for any user seeking personal advice from generative systems.

Regional and Generational Differences in AI Advice‑Seeking

The trend is global but varies by region and age. In the United States, Pew Research data from 2025 indicates that 32% of adults under 30 have used AI for personal advice at least once. In Europe, rates are slightly lower, around 24%, partly due to stronger data privacy concerns. In Asia, particularly in South Korea and Japan, AI companionship apps have been popular for years, and the step to using ChatGPT for life advice is short. Emerging markets like Brazil and India show rapid growth, driven by smartphone penetration and limited access to professional therapy.

Generation Z stands out as the most comfortable with AI advisors. Having grown up with smartphones and recommendation algorithms, they exhibit less hesitation about trusting machine‑generated guidance. Older generations, particularly those over 55, show more scepticism but also higher rates of turning to traditional oracles like astrologers or clergy. The COVID‑19 pandemic accelerated the shift: during lockdowns, many people lost access to in‑person support networks and turned to online tools, including early versions of ChatGPT. That habit has persisted.

One critical observation: Regulatory bodies have taken notice. The European Union’s AI Act, fully in force since 2025, classifies AI systems providing mental health or relationship advice as “high‑risk” if they are used in professional contexts. However, consumer‑facing chatbots like ChatGPT remain largely unregulated for personal use.
This creates a gap: millions of people receive potentially influential life advice from systems with no oversight, no liability, and no requirement to disclose their limitations.

The April 2026 study on tarot practitioners and AI reveals a quiet revolution in how humans seek guidance. Up to 87% of users already consult generative AI for personal advice, often preferring it over human oracles or friends. The reasons are clear: AI is always available, never judgmental, and produces confident answers even to ambiguous questions. But that confidence is a double‑edged sword. When a system hallucinates meaning or reinforces a user’s existing bias, the risk of dependency and distorted decision‑making grows.
The healthiest approach, as some tarot readers have discovered, is to treat AI not as an oracle but as a sparring partner – a tool to challenge one’s own assumptions rather than outsource one’s judgment.

By Jake Sullivan
May 22, 2026

Join us. Our Telegram: @forexturnkey
All to the point, no ads. A channel that doesn't tire you out, but pumps you up.