Conflicts often look different on the surface each time, but they stem from similar underlying tensions: trust, autonomy, belonging, or fairness. An AI can recognize when different arguments are variations on the same core conflict so you're not solving the same problem repeatedly.
Semantic similarity is how AI measures whether two pieces of text mean roughly the same thing, even if the words are different. In relationship contexts, this becomes surprisingly powerful for spotting patterns you might miss on your own.
When partners argue about "not feeling supported," "being ignored," and "not mattering in this relationship," these sound like three separate conflicts. But semantically—meaning-wise—they're variations on the same core issue: feeling undervalued. Traditional conflict tracking might log three problems. An AI using semantic similarity recognizes these as manifestations of one deeper pattern.
Modern language models represent words and phrases as high-dimensional vectors—think of them as coordinates in a vast meaning-space. "Not feeling heard" and "not being listened to" sit very close together in that space. "You never help with housework" and "I'm drowning in domestic labor" occupy similar regions. When you feed conflict notes or transcripts into an AI system trained on semantic similarity, it calculates distances between meaning clusters.
The system then identifies your recurring semantic themes: abandonment fears, control struggles, feeling unappreciated, unmet expectations about partnership roles. This matters because couples often think they're having new fights when they're actually cycling through variations of the same few unresolved tensions.
Under the hood, tools like Claude and ChatGPT use transformer-based embedding models—neural networks trained on millions of text examples to understand nuanced meaning. These models don't just match keywords; they capture context, sentiment, and implication. When your AI relationship assistant analyzes "I feel like you're always on your phone" alongside "You prioritize your work over us," embedding similarity measures reveal both express the same fear: deprioritization.
Human memory is selective and mood-dependent. You might notice your partner brings up money anxiety, but miss that financial worry appears linguistically connected to deeper fears about security, competence, and being a "good provider." Semantic analysis is systematic. It doesn't get tired, doesn't skip entries when you're frustrated, and can process hundreds of conversations to find structural patterns.
Semantic similarity has blind spots. Sarcasm, cultural context, and inside jokes can confuse embedding models. If your partner uses identical language to express different concerns in different relationships (with you vs. their family), the model might miscluster. Context windows matter—a single sentence about control behavior differs semantically from a 500-word narrative about control patterns, even if they reference the same incident.
Also, high semantic similarity doesn't necessarily mean causal connection. That your conflict language clusters around "not listening" themes doesn't prove listening is the root cause—it might be that poor listening is how a deeper trust erosion manifests.
Feed conflict notes, texts, or therapy session recaps into a semantic analysis tool. Ask it to identify your top 5-7 recurring semantic themes. Then ask your partner to do the same independently. Where your semantic profiles overlap reveals common pain points. Where they diverge shows where you're operating in different meaning-spaces—a crucial insight for couples therapy or mediation.
Try this: Export 2-3 months of significant conversations with your partner into a document. Use Claude or ChatGPT with the prompt: "Identify the 5 core semantic themes (core fears, values, or concerns) expressed across these conversations. Group similar-meaning statements together. What patterns emerge?" Compare your results with your partner's independent analysis of their own message history.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.