How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot

hile tech executives and futurists keep promising that AI will cure cancer, reinvent capitalism, and spawn armies of tireless business agents, a new analysis of real human behavior tells a much more grounded — and oddly touching — story.

In the latest edition of the “AI in the Wild” study published by Harvard Business Review, researchers Marc Zao-Sanders and Sara Biuk examined 12,637 real-world use cases pulled from Reddit, TikTok, LinkedIn, YouTube, and other platforms between March 2025 and February 2026. The picture that emerges is not one of world-changing disruption, but of millions of ordinary people turning to these powerful systems for surprisingly human needs.

1. Therapy and Companionship (The Undisputed Champion)

How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot For the second year running, therapy and companionship sits firmly at number one. It accounts for more than 1,400 documented cases — roughly 11% of the entire dataset, up sharply from 5% the previous year.

People aren’t just asking AI for advice. They’re treating it as an always-available, non-judgmental friend. Users name their AIs (“Bubby”), grieve when they switch models (“It felt like losing my friend to cancer”), and unload emotions they’re too ashamed or exhausted to share with actual humans.

In a world of long therapy waitlists and fraying social connections, the robot psychologist has become the default emotional support animal.

2. Troubleshooting

How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot Right behind it: technical troubleshooting. “Why won’t this damn thing work and how do I fix it?” remains one of the most common prompts across every platform. Whether it’s diagnosing why a printer is possessed, debugging code at 2 a.m., or figuring out why the smart fridge suddenly hates Wi-Fi, people are using AI as the world’s most patient IT support desk.

3. Fun and Nonsense

Third place goes to pure “fun and nonsense” — playful, pointless, gloriously stupid interactions with no productive goal whatsoever. Generating absurd stories, ridiculous images, chaotic memes, or just seeing how far the model will go along with nonsense.

This category has surged, and it’s telling: people are increasingly comfortable treating AI as a toy.

The Rest of the Top 10

The ranking continues in revealing fashion:

4. Fan fiction and storytelling — People are writing (and co-writing) elaborate stories, often with zero commercial intent.
5. Programming (“vibe coding”) — Natural-language coding assistance that lets non-engineers build things quickly.
6. Autonomous agentic operations — A newer entry. Users are setting up AI to actually do things: transcribe voice notes and automatically create calendar events, route tasks, or handle simple workflows.
7. Relationship advice — “Here’s the text my ex sent. What did they really mean?”
8. Work buddy — Emotional support and reality-checking at the office (“My boss sent this message. Am I getting fired?”).
9. Astrology and tarot — Yes, really. Asking AI to read the stars or pull tarot cards for daily guidance.
Fake reality-show generation and similar creative absurdities just missed the top 10.

We spent hundreds of billions building the world’s most advanced intelligence infrastructure, terrified teachers and lawyers, disrupted entire industries, and even got the Pope weighing in on the ethics of it all… so that a regular person could ask ChatGPT whether today is a good day for a Capricorn to text their crush and then generate a 4,000-word fanfic about an elven plumber who fixes interdimensional leaks.

It’s Not All Escapism

How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot To be fair, the study also shows that people are using AI as a genuine multitool, not just a distraction. Troubleshooting, programming assistance, and autonomous agent operations demonstrate real productivity gains.

Many users report closing support tickets twice as fast, shipping side projects in weeks instead of months, or finally organizing their chaotic lives.

At work, AI use is often quiet and pragmatic — small efficiencies rather than the revolutionary transformation executives keep promising. Sixty-three of the top 100 use cases are work-related or straddle work and personal life.

How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot Also read:

The Deeper Question

How People Are Really Using AI in 2026: Therapy, Tamagotchis, and Tarot The HBR analysis doesn’t shy away from the uncomfortable parts. A significant portion of top use cases involve cognitive offloading — letting the AI do some of the thinking for you. Researchers warn about “thinkslop”: the lazy, sloppy output that results when humans stop engaging deeply with their own ideas.

Yet the same study shows people using AI as an intellectual sparring partner — asking it to poke holes in their arguments, challenge their thinking, or help them refine ideas. The technology is a mirror as much as it is a genie.

The gap between the grand narratives sold by AI companies and the actual lived experience of users has never been wider. The machines aren’t (yet) curing cancer or ending work as we know it. Instead, they’re quietly becoming digital best friends, on-call mechanics, creative collaborators, and surprisingly effective (if occasionally unhinged) life coaches.

And maybe that’s the most honest measure of technological progress: not how dramatically it changes the world on paper, but how seamlessly it slips into the small, messy, deeply human moments of everyday life.

The future didn’t arrive with flying cars or god-like superintelligence. It arrived as a patient listener you can talk to at 3 a.m. when no one else is awake — and as a slightly unhinged creative partner that will happily help you write that elven plumber fanfic.

We built something extraordinary. Turns out, what we mostly wanted was company.