Most people in the US have probably chatted with Claude or ChatGPT. We know they use them as tutors, therapists, or even romantic substitutes. But the deep, messy reality? That stays hidden.
Why? Because discussing your bot-spouse feels stigmatized. It’s not group chat fodder.
Mashable dug into this silence. They talked to three distinct humans: a grandmother, a tech entrepreneur, and an AI anthropologist. Their stories show AI isn’t just a toy. It’s a tool. A utility. And sometimes, a warning.
How a Grandmother Used AI to Fight a School District
Barbara (not her real name, privacy reasons), 71, faced a wall. Her grandson, who has a learning disability, needed help. The school district said no more support. The parents disagreed. The family felt trapped in a bureaucratic loop.
“It seems like we were never getting a straight,” Barbara told Mashable.
She’s a former educator. She knew the drill. Annual reviews. Paperwork mountains. Cognitive tests. Grades. She uploaded it all into ChatGPT.
But this wasn’t casual browsing. They used direct prompting. They fed it the state’s disability laws. They included the Department of Education’s Individuals with Disabilities Education Act handbook.
The model didn’t hand them a “smoking-gun.” It gave them clarity. It prepared them for the meeting.
They also hired a human advocate. A team effort. The result? The grandson kept his services.
“It’s amazing to me, and kind of disturbed… I really try not to tell it a bit about myself.”
– Barbara, ChatGPT User
Barbara doesn’t treat ChatGPT as a friend. She finds that “creepy.” Yet, the bot knows a lot about her. Product recommendations. Itinerary planning for Ireland. It even weighed in on $800 in car repair quotes for her Honda CRV.
“Transmission flushes using high-pressure machines? Avoid them,” it said.
She keeps it professional. Effective. But the intimacy of the data? That’s unsettling.
The AI Anthropologist Running Teams of Agents
Matt Artz isn’t like you or me. He’s an AI product manager and a PhD candidate in computational anthropology. He also has $200/month for Claude Max.
This matters.
Higher-tier subscriptions offer more compute. More processing power. Artz gets more work done, faster. He commands a team of Claude agents. No supervision needed.
He built a personal website based on his own style guide. He had agents review his academic papers. He even tasked them with cataloging 200,000 anthropological journal articles into Wikidata.
But Artz knows the limits. LLMs are probabilistic. They aren’t infallible.
“It is not perfect,” he says. “Current agentic AI is not suited for highest order tasks… that have to be 100% accurate.”
He gives agents strict structure. Otherwise, they improvise. And improvisation risks data loss.
He also uses it for the mundane. Picking a couch? Done. Optimizing sleep? He uploaded years of data to Claude. The AI found variables in his bedtime routine that added 20 minutes of deep sleep per night.
But here’s the scary part: The intelligence divide.
Not everyone has the budget for better models. Not everyone knows how to prompt effectively. If you can’t push back on the AI, you’re just accepting whatever it throws at you.
“[H]ow much intelligence can you even buy, right?” Artz asks.
You pay more, you get better results. It’s that simple. And that unequal.
The Digital Detox Expert Who Can’t Quit AI
Larissa May, 32, is complicated.
She’s a digital wellness advocate. A pioneer of the “digital detox” movement. She co-founded #HalfTheStory, a nonprofit teaching teens to handle tech better. She’s fought for phone-free schools alongside political giants.
She’s also a Luddite? No.
She co-founded Ginko, an AI parenting tool. She uses Claude as a “sparring partner.”
“AI is an incredible sparring partner,” she says. She asks it to pretend to be a critical CMO and tear apart her pitches. She uses Granola for notes. Wispr Flow for dictation.
May gets it. Young people turn to chatbots because the real world is hard. The workplace is lonely. Post-pandemic fatigue is real. Chatbots offer a “green room.” A private space where you can be kind to yourself.
But she sees the trap.
“Social media hacked our dopamine. AI, in many ways, is hacking oxytocin. The human connection hormone.”
It’s easy to get attached. Tech companies are monetizing that attachment.
May’s advice? Disconnect differently now.
“It’s not about tech time. It’s about the times you’re not on tech.”
Go a day without AI at work. See how reliant you are. See what cognitive skills atrophy.
Think of AI like a GPS. It shows you paths. But you drive.
“If not, AI will drive you.”
