The technology that was supposed to write your emails is now helping people find meaning in their lives. That should make us ask harder questions.
A year ago, generative AI was mostly seen as a productivity hack. Write faster. Code cleaner. Make a spreadsheet less painful. But in 2025, the script has flipped, and fast. According to the latest Top 100 Gen AI Use Case Report, therapy and companionship now top the list of real-world applications.
Not coding. Not copywriting. Not deepfakes or even design.
Talking. Reflecting. Trying to feel better.
What People Are Actually Doing with AI
Let’s get specific. The 2025 report was built from thousands of online posts and user quotes, scouring Reddit, Quora, and niche forums—places where people don’t sugarcoat. And what emerged is striking: AI is no longer just a tool. It’s becoming a mirror. A confidant. Sometimes, even a lifeline.
Here’s the current top five use cases:
- Therapy and Companionship
- Organizing My Life
- Finding Purpose
- Enhanced Learning
- Generating Code (for pros)
That shift—from technical to emotional- isn’t subtle. It reflects a broader truth: the more accessible the tech becomes, the more it wraps itself around our most personal vulnerabilities. People are using Gen AI to get through depression, grief, and burnout. To stay focused. To argue less with themselves. One quote from South Africa stood out:
“There’s a psychologist for 1 in every 100,000 people and a psychiatrist for 1 in every 300,000. Large language models are accessible to everyone. Unfortunately, data safety is not a concern when your health is deteriorating and survival is the morning agenda.”
That’s not just an anecdote. That’s infrastructure failing, and AI filling in the cracks.
The Emotional Algorithm: What’s Behind the Shift?
There are reasons why Gen AI is filling this role now. Let’s unpack a few.
1. Always On, Never Judging
Human therapy is expensive, stigmatized, and often hard to schedule. Gen AI, by contrast, doesn’t roll its eyes, doesn’t run out the clock, and doesn’t cost $200 per hour. It’s just… there.
2. Cost Has Dropped. Access Has Exploded.
In the past 12 months, usage has doubled. Major players like OpenAI, Google, and Microsoft have pushed AI into every corner, from mobile apps to car dashboards. Prices are down. Voice interfaces are up. Custom GPTs and agentic systems are spreading fast.
3. It Works—Well Enough
Multiple studies now suggest that AI-generated therapeutic responses are often indistinguishable from those written by trained psychologists. Not always better, not always right, but good enough to keep people coming back.
Self-Help 2.0: AI as Life Coach, Executive Assistant, and Spiritual Guide
Another surprise: “Organizing my life” came in at #2 this year. This isn’t just digital calendars—it’s AI helping people with New Year’s resolutions, meal planning, and even getting out of bed. Micro-changes. Momentum shifts. One user put it simply:
“I asked it to build a timeline for me to clean and organize my house before guests arrived. I just followed it.”
And then there’s #3: “Finding purpose.” What used to be the domain of existential philosophers, TED Talks, and long walks at 2 a.m. is now… AI-enhanced.
It sounds absurd until you try it.
People are asking questions like:
- “What should I do with my life?”
- “How do I stop procrastinating?”
- “Why do I keep sabotaging myself?”
And they’re not getting empty aphorisms, they’re getting frameworks, nudges, analogies, even rituals. Not always perfect. But sometimes enough.
Meanwhile, the Experts Are Watching Closely
At Microsoft, Jared Spataro called Gen AI “a personal assistant at work,” one that connects to your files, emails, and calendar to take over the drudgery and make room for creative, strategic thinking.
At EY, they’re training 400,000 employees to work with “agentic” AI systems, including 150 agents deployed for tax-related tasks alone.
And yet the emotional edge keeps growing. AI isn’t just a tool. It’s becoming a presence, and that’s something most experts didn’t predict.
As Allie Miller, a Fortune 500 AI advisor, put it:
“The lack of judgment and unrestricted exploration makes it an ideal playground for big dreams, potentially embarrassing questions, or hazy, half-formed goals.”
It’s a strange place to be: trusting a machine more than a mentor. But here we are.
The Backlash Is Real—and Worth Listening To
Let’s be honest. Not everyone’s thrilled about this.
Some users admit they’re already too reliant. One said, “Rather than use my brain for a complex task, I’m just turning to GPT.” Others worry about what happens to children, students, or anyone still learning how to think critically when a robot is always standing by with an answer.
Add privacy concerns to the pile. Reddit’s full of posts like this:
“Too late: my bank has it, my credit cards have it, Google has it… texts, emails, voicemails, location, food orders, sleep data. You name it.”
And ironically, some users are now complaining that LLMs don’t know enough about them—that their memory resets are too aggressive, that personalization is too shallow.
So which is it? Too much memory? Or not enough?
That tension—between what AI knows and what it should know, is going to shape the next phase of this conversation.
Thinking Ahead: From “Telling” to “Doing”
One major theme emerging from forums this year is agency. People don’t just want AI to advise. They want it to act.
“I want it to cancel this subscription before they start charging me.”
Right now, LLMs are largely static. They generate language. But in the near future, we’ll see more agentic AI models that not only recommend actions but take them on your behalf, with or without a human in the loop.
That changes the game. And it forces a tougher reckoning: How much of your decision-making are you willing to outsource?
So, What Do We Make of All This?
There are two ways to see what’s happening.
One: AI is simply filling in where institutions and relationships have failed. Therapy gaps, educational holes, work that never ends, AI just steps in. It’s not deep. It’s not scary. It’s just triage.
Two: We are slowly rewiring what it means to think, feel, and relate, with a machine as our co-pilot. And the longer we ride with it, the harder it will be to go back.
The truth probably lives between those two extremes.
But either way, 2025 has made one thing very clear: AI’s most powerful feature isn’t its ability to finish your sentence. It’s that it’s always listening, always answering, and never tells you you’re being too much.
And that, for better or worse, is changing us.
Sources:
- Top 100 Gen AI Use Cases – 2025 Report
- Allie Miller, Fortune 500 AI Advisor
- Jared Spataro, Microsoft
- EY Global L&D reports
- Reddit, Quora, Filtered.com data
