Maggie returned to the lab three days later, having told Tom she was working late. The lie came easily, as lies often did when they were small. But now, looking at the screen, she wondered: how many small lies had she told? How many had the AI captured? Dr. Chen met her in the same small room, the same white walls, the same hum of servers. "The deep analysis is complete," he said. "Are you ready to see it?" Maggie sat down at the terminal, her stomach tight with something that felt uncomfortably like the anxiety she saw in her clients before a verdict. "Show me." The screen filled with data--graphs, charts, timelines, each one a window into her life over the past two years. Chen had organized it into categories: Communication Patterns, Shared Activities, Emotional Indicators, Conflict Resolution, Future Planning. "Let's start with communication," Chen said, clicking on the first category. "The preliminary analysis showed declining frequency. The deep analysis reveals something more interesting." A new graph appeared, showing the content of her communications with Tom--not the actual words, but the topics the AI had identified through pattern recognition. The categories were labeled: Logistics (household management, schedules, errands), Social (friends, family, events), Work (career discussions, professional stress), Personal (feelings, dreams, fears), and Conflict (disagreements, tensions, unresolved issues). "Your logistics conversations have remained stable," Chen explained. "About 60% of your communication with your spouse falls into this category--discussing schedules, managing the household, coordinating activities. This is normal for long-term couples." Maggie nodded. They'd always been efficient. Two busy professionals managing a life together. "But look at the personal category." Chen pointed to a thin line at the bottom of the graph. "Two years ago, personal conversations made up about 25% of your communication. Now it's less than 8%." Maggie stared at the line. She remembered when she and Tom used to talk for hours--about their dreams, their fears, the things that mattered. When had that stopped? When had they become two people who only discussed logistics? "The system also tracks what it calls 'emotional language'--words and phrases that indicate vulnerability, intimacy, or connection." Chen pulled up another chart. "Your use of emotional language with your spouse has declined by 61% over two years." "That doesn't mean our marriage is in trouble. It means we're comfortable. We don't need to constantly express our feelings." "Comfort is one explanation. The system also considers another: emotional withdrawal." Chen's voice was gentle, but the words landed like a blow. "When couples stop sharing their inner lives, they often stop seeing each other as confidants. They become... roommates. Functional partners. But not intimate ones." Maggie felt a flash of anger. "An algorithm can't understand intimacy. It can't capture what happens between two people." "You're right. The system can only see patterns. It can't see meaning." Chen paused. "But the patterns are consistent with what relationship researchers have observed in couples approaching separation. The withdrawal happens slowly, incrementally. So slowly that the people involved often don't notice until it's too late." The next category was Shared Activities. Maggie had expected this one to be better--they did things together, didn't they? The data told a different story. "Two years ago, you and your spouse engaged in shared leisure activities an average of 4.2 times per week," Chen said. "Dinners out, movies, walks, hobbies--activities you did together, without work or other obligations." "And now?" "1.7 times per week. A decline of 60%." Maggie thought about their weekends. Tom working on his architectural projects, her catching up on case files. The occasional dinner out, but even those had become perfunctory--quick meals between other commitments, not the leisurely experiences they used to be. "The system also tracks what it calls 'parallel presence'--times when you're physically together but engaged in separate activities." Chen showed her another graph. "This has increased by 43%. You're in the same space, but not truly together." "We're both busy. We have demanding careers." "True. But the system has data from other high-achieving couples. The ones with strong marriages maintain shared activities even during busy periods. They prioritize connection over efficiency." Chen looked at her carefully. "Your data suggests you've stopped making that choice." Maggie felt the anger rising again, but this time it was mixed with something else. Fear? Recognition? She wasn't sure. "What else?" she asked, her voice tight. The Emotional Indicators category was the most disturbing. The AI had analyzed her communications--not just with Tom, but with everyone--and identified patterns in her emotional expression. "Your overall emotional vocabulary has declined," Chen said. "Not just with your spouse, but across all relationships. You use fewer words related to feelings, more words related to tasks and obligations." "I'm a lawyer. We're trained to be precise." "This isn't about professional communication. The system separates work from personal contexts. Even in your personal communications, you've become more... functional. Less expressive." Maggie stared at the data. It was true--she'd noticed it herself, sometimes. The way she struggled to express what she was feeling. The way she defaulted to practical concerns instead of emotional ones. She'd attributed it to maturity, to professional development, to the natural evolution of a focused mind. But what if it was something else? What if she was slowly shutting down? "The system also tracks what it calls 'avoidance patterns'--behaviors that indicate emotional avoidance." Chen showed her a timeline of her activities. "Your work hours have increased by 23% over two years. Your time spent on hobbies has decreased by 45%. Your social engagements outside of work have decreased by 38%." "I've been building my practice. That takes time." "Of course. But the system also shows that you've been avoiding situations that might require emotional engagement. You've declined 67% of social invitations that involve couples or intimate settings. You've stopped attending family gatherings, citing work conflicts. You've even reduced your communication with your sister, who you used to speak with weekly." Maggie felt a chill. She hadn't realized the AI was tracking all of that. She hadn't realized she'd been doing it. "I'm an introvert. I need downtime." "The system accounts for personality type. This isn't introversion. This is withdrawal." Chen's voice was soft. "You're pulling back from emotional connections, Maggie. Not just with your husband, but with everyone. And you may not even realize you're doing it." The Conflict Resolution category was brief but pointed. The AI had identified only three significant conflicts between Maggie and Tom over the two-year period--all minor disagreements that had been resolved quickly. "That's good, isn't it? We don't fight." "The system doesn't see it as good." Chen pulled up a comparison chart. "Healthy couples have conflicts. Not constant fighting, but regular disagreements that they work through together. The process of resolving conflict builds intimacy. It requires communication, compromise, emotional engagement." "And we don't have that?" "You have avoidance. When the system detects potential conflict points--moments where disagreement might occur--it shows both you and your spouse backing away. Changing the subject. Deflecting with humor or distraction. You're not resolving conflicts. You're preventing them from happening." Maggie thought about the times she'd bitten her tongue, let things slide, chosen peace over honesty. She'd thought she was being a good wife. Was she actually being a distant one? "The absence of conflict isn't harmony," Chen said. "It's disengagement. Couples who've given up on their relationship often stop fighting. They stop expecting anything to change." The final category was Future Planning. The AI had analyzed her communications for references to the future--plans, dreams, goals, expectations. "Two years ago, 34% of your communications with your spouse included references to shared future events--vacations, home improvements, retirement plans. Now it's 12%." "We're busy. We focus on the present." "The system also shows that your individual future planning has increased. You talk about your career, your goals, your plans--but increasingly, you talk about them as an individual, not as part of a couple." Maggie stared at the screen. The data was overwhelming, a cascade of patterns she'd never noticed, never questioned. The AI had captured something she'd been hiding from herself: a marriage that was slowly, quietly dying. "The prediction is ready," Chen said quietly. Maggie looked at him, her lawyer's composure cracking. "Show me." He typed a command, and a new screen appeared. At the top, in large numbers, was a single statistic: Divorce Probability: 78% Below it, a timeline: Estimated time to separation: 3-5 years And then, a list of contributing factors: - Communication decline (high impact) - Emotional withdrawal (high impact) - Parallel living patterns (moderate impact) - Conflict avoidance (moderate impact) - Future planning divergence (moderate impact) Maggie stared at the screen, her mind racing. This couldn't be right. She knew her marriage. She knew herself. An algorithm couldn't possibly understand the complexity of her life. But the data was there. The patterns were clear. Deep inside, she felt a recognition she couldn't deny. "I need to go," she said, standing abruptly. "Ms. Sullivan--" "I need to think about this." She gathered her things, her hands shaking slightly. "I'll be in touch." She walked out of the lab into the October afternoon, the prediction following her like a shadow. Seventy-eight percent. Three to five years. The numbers echoed in her head, a verdict she hadn't expected, a truth she wasn't ready to face. Her phone buzzed. Tom: Everything okay? You seem distant lately. She stared at the message, the irony not lost on her. The AI had seen what she hadn't. The distance. The withdrawal. The slow erosion of a marriage she'd thought was fine. She typed back: Just busy. See you at dinner. Another lie. Another pattern the AI would capture. Another data point in the analysis of a marriage that, according to the algorithm, was slowly ending.
The number appeared on the screen like a verdict. 78%. Probability of divorce within five years. Maggie stared at it, her mind refusing to process what she was seeing. "That's impossible," she said. "My marriage is fine." Dr. Chen sat across from her, his expression carefully neutral. "The system doesn't make judgments about whether a marriage is 'fine.' It identifies patterns and predicts outcomes based on those patterns." "Then your patterns are wrong." Maggie stood up, pacing the small room. "Tom and I have been together for eighteen years. We have a good life. A good partnership. We're not, we're not one of your data points." "Actually, you are. That's how the system works." Chen's voice remained calm. "It compares your behavioral patterns to thousands of other couples and identifies similarities. The couples whose patterns match yours have divorced within five years in 78% of cases." "Then the other couples are different. Their situations are different." Maggie felt her lawyer's instincts kicking in, the need to argue, to find the flaw in the evidence, to dismantle the opposing case. "You can't reduce human relationships to algorithms. Every marriage is unique." "That's true. And the system accounts for uniqueness by using multiple variables and adjusting for individual circumstances. But the patterns it identifies, communication decline, emotional withdrawal, parallel living, these are robust predictors across diverse populations." Maggie stopped pacing and faced him. "What if I don't accept the prediction?" "Many people don't. That's part of what we're studying." Chen gestured to the screen. "The system shows that people who reject predictions often do so because the predictions conflict with their self-image. They believe they know themselves better than any algorithm can." "And they're wrong?" "Sometimes. Sometimes they're right. The system isn't infallible, no prediction model is. But the accuracy rate is high enough that dismissing it entirely might be unwise." Maggie felt a flash of anger. "You're saying I should believe a machine over my own experience?" "I'm saying you might consider whether your experience is giving you the full picture." Chen stood up, walking to the screen. "Look at this data. Communication down 12%. Personal conversations down 47%. Shared activities down 60%. These aren't the system's opinions. These are your behaviors, captured and analyzed. The question isn't whether the machine is right. The question is whether you've been paying attention." --- Maggie left the lab in a fog of denial and determination. The AI was wrong. It had to be. She knew her marriage better than any algorithm. She drove home through the Chicago traffic, her mind churning through counterarguments. The system didn't account for context. It didn't know about the pressures of her career, the demands of running a law firm, the natural evolution of long-term relationships. It didn't understand that marriages changed, that passion evolved into partnership, that comfort wasn't the same as decline. Tom was in the kitchen when she arrived, chopping vegetables for dinner. He looked up as she entered, his expression warm. "Hey. How was your day?" "Fine. Busy." Maggie hung up her coat, forcing a smile. "What are you making?" "That stir-fry you like. With the cashews." He returned to his chopping, the knife moving in a familiar rhythm. "I was thinking we could watch that new documentary tonight. The one about the architect you mentioned?" Maggie paused. The one about the architect. She'd mentioned it weeks ago, and he'd remembered. That was the kind of thing a good husband did. A good partner. "Sounds great," she said. She went to change out of her work clothes, but instead of heading to the bedroom, she found herself in her home office, pulling up the data on her personal tablet. She'd downloaded a copy before leaving the lab, a violation of the study protocol, probably, but she needed to examine it more closely. The numbers stared back at her, implacable. Communication decline. Emotional withdrawal. Parallel living. Each category was broken down into subcategories, each subcategory supported by specific data points. She clicked on Communication Patterns, looking for flaws. The system tracked phone calls, texts, emails, and in-person conversations detected through smart home voice recognition. But it couldn't capture everything. It couldn't capture the glances across the dinner table, the inside jokes, the comfortable silences that meant intimacy rather than distance. Could it? She thought about dinner the night before. Tom had cooked. She'd checked her email. They'd talked about his project at work, about a scheduling conflict with the car maintenance, about whether they needed to order more dog food. Functional conversations. Logistics. When had they last talked about something that mattered? She tried to remember. A conversation about dreams, about fears, about the future they were building together. She drew a blank, not because there hadn't been such conversations, but because she couldn't pinpoint when they'd occurred. That was normal, wasn't it? Long-term couples didn't need to constantly reaffirm their connection. They knew each other. They were secure. But the AI had called it something else: emotional withdrawal. --- Over the next week, Maggie became obsessed with proving the prediction wrong. She approached her marriage like a case, gathering evidence, building arguments, looking for the flaw in the prosecution's theory. She started paying attention to her interactions with Tom, not through the lens of the AI, but through her own observations. She noted the times they laughed together, the small gestures of affection, the moments of connection that no algorithm could capture. Tuesday morning: Tom brought her coffee in bed, the way he always did on days when she had early meetings. That was love, wasn't it? That was partnership. Wednesday evening: They watched the documentary about the architect, sitting together on the couch, his arm around her shoulders. They discussed it afterward, the design principles, the way architecture shaped human experience. That was intellectual connection. That was shared interest. Thursday night: They had dinner at their favorite Italian restaurant, the one they'd been going to for years. The owner knew them by name, brought them their usual wine without being asked. That was history. That was commitment. Each moment felt like evidence. Each interaction was proof that the AI was wrong, that her marriage was fine, that she knew herself better than any machine. But even as she gathered this evidence, she noticed something else. The moments of connection were fewer than she'd remembered. The comfortable silences sometimes felt like gaps. The conversations about logistics far outnumbered the conversations about feelings. And she was still checking her email during dinner. On Friday, she returned to the lab. "I want to see the raw data," she told Chen. "Everything the system captured. I want to understand how it reached this conclusion." Chen nodded, unsurprised. "That's reasonable. The data is yours, you have a right to examine it." He led her to a private terminal and pulled up the complete dataset. Thousands of data points, organized chronologically, each one a small piece of her life over the past two years. Maggie scrolled through the records, her lawyer's eye catching details. Phone calls: duration, timing, recipient. Texts: frequency, response time, word count. Emails: sender, recipient, subject line. Smart home data: room occupancy, voice detection, activity levels. Financial transactions: merchant, amount, date. Calendar entries: appointments, meetings, events. It was overwhelming. A life reduced to data points, each one meaningless on its own but collectively forming a picture she couldn't dismiss. "Show me the communication analysis again," she said. "The part about emotional language." Chen pulled up the relevant section. The system had identified keywords and phrases associated with emotional expression, words like "feel," "wish," "hope," "fear," "love", and tracked their frequency across all her communications. "Your use of emotional language has declined across all contexts," Chen explained. "But the decline is most pronounced in your communications with your spouse. Two years ago, emotional words made up about 8% of your communications with him. Now it's less than 2%." "That doesn't mean I don't love him. It means I express it differently." "Perhaps. But the system also tracks what it calls 'affirmative language', expressions of appreciation, gratitude, affection. These have declined by 71%." Maggie stared at the number. Seventy-one percent. She tried to remember the last time she'd told Tom she loved him, not as a perfunctory sign-off to a text, but as a genuine expression of feeling. She couldn't. "What about him?" she asked. "Does the system track his communication patterns too?" "We only have access to your data. But we can infer some patterns from your interactions. For example, the system shows that he initiates communication more often than you do, and that his initiation rate has remained stable while yours has declined." "So he's trying. I'm the one pulling away." "That's one interpretation. The system would say that both parties are contributing to the pattern, you through withdrawal, him through acceptance of the withdrawal." Maggie felt a chill. She thought about Tom, about his patience, his warmth, his consistent presence. She'd always appreciated those qualities. But what if his patience was actually resignation? What if his warmth was a gift she'd stopped reciprocating? She left the lab with a new determination. The AI was wrong. It had to be. She'd prove it by changing the patterns, by reconnecting with Tom, by showing that self-knowledge was still possible. That night, she came home early, a rarity, and found Tom in his home office, working on a design project. "Hey," she said, leaning in the doorway. "Got a minute?" He looked up, surprised. "Sure. Everything okay?" "Everything's fine. I just..." She paused, unsure how to begin. The AI had shown her the patterns, but it hadn't given her a script for changing them. "I was thinking we could talk. Really talk. Not about logistics or schedules. Just... talk." Tom studied her for a moment, his expression unreadable. Then he smiled, but there was something sad in it. "Okay. What do you want to talk about?" Maggie sat down across from him, suddenly aware of how long it had been since they'd done this, just sat together, without an agenda, without distractions. The silence stretched between them, and she realized she didn't know where to start. "How are you?" she asked finally. "Really. How are you?" Tom's eyebrows rose slightly. "I'm fine. Work is busy. The Henderson project is finally moving forward." "That's great. How do you feel about it?" He paused, as if the question were unexpected. "Good. It's a good project. I'm glad to be working on it." Maggie nodded, but she could feel the conversation stalling. She was trying, really trying, to connect. But the patterns were deeply ingrained, and she didn't know how to break them. "What about us?" she asked. "How do you feel about us?" Tom was quiet for a long moment. Then he said, carefully, "I think we're okay. We're both busy. We're both focused on our careers. That's normal, isn't it?" The words echoed the AI's analysis: parallel living, emotional withdrawal, acceptance of distance. Maggie felt a flash of panic. Was this what the algorithm had seen? Two people who had settled for okay? "I want more than okay," she said. Tom looked at her, really looked at her, for the first time in what felt like years. "So do I," he said quietly. "But I didn't think you did." The words hit her like a physical blow. She opened her mouth to deny it, to argue, to prove him wrong. But the data was there, in her mind, impossible to ignore. Communication decline. Emotional withdrawal. A marriage slowly dying because she'd stopped feeding it. "I didn't know," she said, her voice barely a whisper. "I didn't realize I was doing it." Tom reached across the space between them and took her hand. "I know," he said. "I don't think you meant to." Maggie felt tears prick at her eyes, tears she hadn't shed in years, tears the AI had probably predicted based on some pattern she didn't understand. The prediction was 78%. But in this moment, holding her husband's hand, facing the truth she'd been avoiding, she realized that the number wasn't a verdict. It was a warning. And warnings could be heeded.