If you’ve ever turned to artificial intelligence to try to figure out how to handle a tricky situation with a friend or colleague, you’re far from alone. For many, AI has become a modern oracle – a source of guidance, emotional support or clarity in moments of uncertainty – though critics worry that they could lead to emotional dependence on the technology.

Of course, the urge to seek answers from forces beyond ourselves is hardly new. For generations, people have turned to psychics, astrology charts or tarot cards for reassurance.

Once fringe, these practices have increasingly become mainstream. According to a 2025 Pew Research survey, nearly 1 in 3 Americans consult tools such as tarot or astrology at least once a year, interest that’s thought to largely be fueled by Gen Z and social media.

Now, we’re seeing these two forces – AI and occult practices – meeting in strange and fascinating ways. An increasing number of tarot readers, from novices to seasoned practitioners, have been turning to AI to help make sense of their tarot readings.

What makes this pairing so striking is that interpretation is the whole point of tarot. And yet AI often brings little knowledge of your history or your unique situation when it dispenses advice.

In a study published in April 2026, we examined which aspects of the practice that tarot readers were delegating to AI, and how the technology was shaping their interpretations.

Watching what happens when readers hand that important interpretive step to AI may offer a glimpse of what helpful AI guidance could look like – and where it could go wrong.

The mainstreaming of occult practices

Tarot cards are experiencing a revival.

Tarot did not start out as a spiritual or fortune-telling tool. It began as a popular card game in the Italian Renaissance, before spreading across Europe.

Over time, readers and occultists layered the cards with mystical symbolism drawn from Kabbalah, Egyptology, numerology and other mystical and symbolic traditions. In the early 20th century, the British publisher William Rider & Son released the Rider-Waite-Smith deck, which became the most popular tarot deck in the English-speaking world.

Whereas only a handful of tarot decks were being published in the early 1970s, today thousands of tarot and oracle decks are in circulation. A standard tarot deck contains 78 cards, each carrying its own symbolic meaning. Practitioners use the cards to sit with hard questions, which can range from difficult relationships to world events: Should I leave my partner? Is this job worth it? What’s going to happen with Donald Trump and the Strait of Hormuz?

After cards are pulled, their meanings are interpreted through the lens of the reader’s question, circumstances and life history.

Someone asking about a relationship and drawing the Tower card, for instance, might read it as impending rupture, or as false assumptions finally giving way. Which reading fits depends on the other cards, the specific question and what the reader already knows about their own situation.

This stands in contrast to AI, which is primed to produce a seemingly definitive answer, even when it’s unaware of the nuances of your situation and context.

The adoption of AI in tarot reading

For our study, we interviewed 12 tarot practitioners about their use of AI in readings they did for themselves.

They generally found themselves pulled in two directions.

On the one hand, they often sought explicit guidance from AI in the process of self-reflection. By using AI to interpret the cards, they could sidestep the frustration of interpreting many cards in light of the question asked.

Say someone drew the Fool and the Ten of Wands for a question about a career change. The Fool points toward a leap into the unknown, while the Ten of Wands speaks to burnout and an unsustainable load.

But do the cards say, “Leave, you’re exhausted and something better awaits”? Or “Leave, and the new job will be just as demanding”?

Rather than sit with that ambiguity, some readers simply ask the AI for the meaning of the reading.

An attendee at Google’s 2025 I/O developers conference wears Android XR glasses with Gemini AI, which she’s using to interpret a tarot card. Camille Cohen/AFP via Getty Images

For more challenging readings, AI’s “yes man energy” helped them feel more confident about their interpretations. This was true for cases where participants both drew physical tarot cards and then interpreted them with AI, or used AI to directly simulate tarot readings.

These uses of AI are seductive. They make the act of self-reflection less demanding. But within the broader tarot community, we found a lot of criticism of AI, and there were concerns about how the sycophantic nature of the technology could undermine people’s intuition and reasoning.

AI as a tool for critical engagement

On the other hand, the tarot readers we interviewed also used AI as a tool to challenge their own biases and assumptions – blind spots in their readings, or what they might be missing in their own interpretation of the cards.

Along these lines, they used AI to generate alternative perspectives so they could compare the different interpretations and see which resonated more. And some even asked for an “objective reading” of the cards, because AI appears to have no skin in the game and be unburdened by personal biases or motives.

Many readers did this when they didn’t want to “bug” or “pester” their friends for help with a reading. Instead, they relied on chatbots in a one-sided relationship that feels supportive – an example of what scholars call parasocial interaction.

Some interviewees even treated bizarre AI-generated outputs or hallucinations as meaningful precisely because they were random and unintended, the same way that a card drawn at random feels like it carries a secret message.

What does this mean for the future of AI?

AI is becoming a powerful new oracle in its own right.

In one recent survey, researchers found that up to 87% of generative AI users are consulting the technology for “personal applications,” which includes advice and emotional support for relationship conflicts and mental health struggles.

Sometimes these chatbots are genuinely helpful. But at the same time, advice seekers can also become emotionally dependent. Some rely on the technology for companionship and guidance instead of friends and family. Chatbots have also been found to nurture delusional beliefs and even lead to self-harm.

Meanwhile, professionals that regularly give guidance are using AI in their practice, from lawyers to therapists and even priests. Pope Leo XIV recently urged priests to resist the temptation to use AI to write sermons.

We think it’s important to make sure the technology isn’t seen as an all-knowing source of truth. It can certainly open up users to new ideas, but it should be a tool to enhance self-reflection, rather than one that serves as a substitute for it.

In some cases, that’s what the tarot readers in our study did. They tapped into their own capacity for reflection by using AI to explicitly challenge their own biases and assumptions. This points to an alternative blueprint for the future of AI – one in which the technology doesn’t simply hand you answers but keeps you actively engaged in the process of finding them.

Ziv Epstein, Postdoctoral Associate, Schwarzman College of Computing, Massachusetts Institute of Technology (MIT); Farnaz Jahanbakhsh, Assistant Professor of Electrical Engineering and Computer Science, University of Michigan, and Vana Goblot, Lecturer in Media, Communications and Cultural Studies, Goldsmiths, University of London

This article is republished from The Conversation under a Creative Commons license. Read the original article.