The notification popped onto Elijah's laptop screen "You seem stressed. Remember, tomorrow is another day! 🌟"
He gazed at it for a good minute before he realized his own system had just attempted to reassure him. The irony wasn't lost on him; sitting in his vacant office at 3 AM, midst boxes of failed prototypes, receiving a pep talk from the very thing that had ruin his life.
(Inage generated with Google ImageFx)
But let me back up.
Two years ago, Elijah Vasquez thought he was going to save the world. Or at least make it a little less bad. He had spent six years working for Google, watching people become increasingly isolated, increasingly depressed, increasingly disconnected. Social media was supposed to bring us together, but instead it just made everyone feel terrible about themselves.
"What if," he'd told his girlfriend Priya during dinner, "what if we could really make people feel better? Automatically?"
Priya had looked up from the pad thai. "What do you mean?"
"I mean, what if an AI could tell that someone's having a bad day and just fix it? Send them just the right thing to say?"
She'd laughed – she'd thought he was being kind but idiotic. "Eli, you can't measure loneliness by an algorithm."
But he thought he could.
The initial prototype was simple. Facial recognition software able to detect micro-expressions, combined with natural language processing to construct personalized responses. Give it enough input regarding what brought people joy, and presto; instant emotional care.
The first investor meeting had gone better than he'd expected. Three guys from Palo Alto Venture capital firm, in hoodies.
"So you're telling," one of them had said, scrolling through his phone as Elijah pitched, "this thing can detect when I'm sad and then what, send me a joke?"
"Beyond that," Elijah had continued, clicking to the next slide. "It learns your specific emotional patterns. Humor doesn't compute for you. Motivational quotes. Maybe you need a reminder of something pleasant. The system adapts."
The lead investor had only just now lifted her head. "Market size?"
"Mental health technology is a forty-billion-dollar market. But we're not selling to clinically depressed people. We're selling to everyone who's ever had a horrible day."
They'd invested him with 2.5 million to start off.
The initial year was phenomenal. His team of twelve engineers worked round the clock, creating what they called the Empathy Engine. They started with a simple app: users would opt to turn on mood tracking, and the AI would learn their routines and preferences.
But Elijah had grander ideas. Why settle for an app when you could integrate it into everything?
That's when things got weird.
"Mood-sensitive streetlights?" Priya had asked when he'd presented the plan for growth. She lay in bed, laptop spread wide, his mind whirling with ideas.
"Think about it," he'd told me. "Someone is walking home from the office, slouching along, staring at their phone. The streetlight catches the slouch, the stride, the scowl. It changes color to something warm and comforting or plays calming music. It could even speak to them personally."
"That's surveillance."
"It's concern, not surveillance. It's having a friend who always knows best for you."
The second round of funding had twelve million pouring in. Overnight, they had forty staff, a fancy office in SOMA, and deals with three big cities on their "Emotional Infrastructure Initiative."
The kiosks were the best. Thin white terminals with a sleek phone-booth appearance, placed in heavy-use areas; train stations, shopping malls, college campuses. Step up to one with your head down low, and it would scan your face, analyze your voice patterns, and dispense exactly what you needed. Sometimes a printed estimate. Sometimes a small gift; a piece of candy, a stress toy, a packet of seeds to plant.
"It's like a vending machine for emotions," Elijah had told the TechCrunch journalist who had come to interview him. The article called him "The Empathy Entrepreneur" and included a photograph of him standing in front of one of the kiosks, arms crossed, beaming.
They'd placed fifty kiosks in San Francisco by month eighteen. The numbers for usage were staggering; thousands of interactions a day, overwhelmingly positive reviews. Everyone loved it.
And of course, there was the coffee shop rollout.
"AI therapists in every Starbucks," the Business Insider headline had touted. "Startup Promises to Turn Your Barista into a Mental Health Professional."
That wasn't just right, but close enough. The idea was to install emotion-reading cameras and speakers in restaurants, coffee shops, waiting rooms, just anywhere stressed or blue people could happen by. The AI would eavesdrop on conversations, read body language, and pipe up when necessary.
"Excuse me," the AI voice would interject over unobtrusive speakers, "I couldn't help noticing you seem stressed out. Here's a breathing exercise that might help you."
The backlash arrived quickly and energetically.
"This is dystopian as hell," someone had tweeted, posting a video of themselves being lectured unsolicited by a Panera's ceiling. "I just wanted to have a sandwich in peace."
Privacy activists started calling it "emotional surveillance." The ACLU submitted complaints. Three cities dropped out of their contracts overnight.
But Elijah stood firm. They were so close to something revolutionary. People just didn't understand yet.
The third round of financing was tougher. Twenty-eight million, but with strings attached. The investors wanted to see six months' profitability. They wanted the privacy issues solved. They wanted to see evidence that the system actually worked.
That's when cracks started to show.
"The problem," Dr. Lou had explained, in one of their consultation meetings, "is that you're approaching emotions like they're something to be solved. But sadness is not always a bug. Sometimes it's a feature."
Dr. Lou was a behavioral psychologist they'd hired to help perfect the AI's responses. She'd been skeptical from the start, but Elijah'd had his heart set on it.
"I don't understand," he'd said.
"Grief, disappointment, even anger; these are normal human responses. When your system sees someone crying and instinctively tries to make them happy again, it's not helping them with their emotional processing. It's teaching them how to deny them."
"But if we can get people to feel better ----
"Can you, though? Your own statistics show ninety percent of the activity is with the same small group of users. You're not fixing loneliness. You're causing dependency."
The studies started showing up about month twenty. Researchers working independently and keeping track of the kiosks said that regular users were getting worse as the months went by, not better. They were addicted to the external validation, unable to self-regulate in the absence of the AI.
Meanwhile, technical problems were piling up. The facial recognition system was unable to handle people of color. The language processing was unable to handle regional dialects. The system kept misreading usual expressions as signs of distress, leading to mortifying interventions.
"Congratulations on your promotion!" it had told a woman at a funeral, apparently mistaking her black funeral dress for business attire.
"You're looking lonely. Here's a joke!" it'd announced to a man who sat reading a book in a park in silence.
The memes were ruthless. #EmpathyFail trended for weeks.
Priya had departed by month twenty-two.
"I can't do this anymore, Eli," she'd said, stuffing her books into boxes. "You're not the same person. You're so committed to fixing everyone else's emotions that you've forgotten how to feel yours."
"I'm trying to help people."
"No, you're trying to justify you're right. And you're burning it all down to do so."
The final straw occurred when there was a Portland city council meeting. They'd been trying to obtain approval for a larger kiosk program when a young adult got up during public comment.
"I used one of those things a day for three months," she'd explained, her voice shaking. "Whenever I was feeling sad or scared, I'd go down to the kiosk and it would tell me I was fine, that things would be all right. But they never were all right. And I stopped learning how to deal with my problems because I always had this machine dictating to me what I should feel."
She'd glared directly at Elijah, who was in the front row. "I ended up in the hospital after a panic attack because I didn't have any idea how to make myself stop freaking out without your stupid robot."
The council voted unanimously to ban the kiosks.
The company collapsed three months later. Forty employees were laid off. Investors who took their losses on taxes. Patents sold to a data mining company for pennies on the dollar.
And now here was Elijah, alone in his apartment at 3 AM, staring at the screen of his laptop. The message remained: "You seem stressed. Remember, tomorrow is another day! 🌟"
He'd left one prototype running, more out of curiosity than anything else. It had been monitoring his computer usage, his facial expressions from the webcam, probably his typing style. It knew he was depressed, poor, and lonely.
And its response was a dull cliche with a star emoji.
He started to laugh. The sort of laughter you laugh when you've spent two years and fifty million dollars producing something that couldn't even comfort its creator.
