About
Sundeep Verma is a product builder and systems thinker working at the intersection of AI, relational computing, and human interaction design. His work focuses on building AI-native relationship systems that move beyond traditional chat interfaces toward continuity-aware, trajectory-aware interaction models. He has been exploring how concepts such as relational state, adaptive memory, trust drift, reinforcement loops, and behavioral continuity can function as foundational computational primitives for future AI systems. His current work investigates the gap between response generation and long-term relational coherence in AI-human interaction.
Talk details
Why Chat Interfaces Fail Human Relationships
About this talk
Every day, millions of people talk to AI systems that can sound empathetic and intelligent, yet the relationship often breaks the moment continuity is tested. The AI forgets, shifts tone, loses context, or responds without awareness of history or emotional trajectory. This talk explores why today’s chat interfaces fail at human relationships, and why the future of AI may depend not on better responses, but on designing systems that can sustain continuity, memory, and and relational under.
Key takeaway
- 1. Understand why response quality alone is insufficient for building long-term human-AI trust, and why continuity failures damage user perception more than isolated weak outputs.
- 2. Learn the difference between conversational memory and relational state, including why memory retrieval alone cannot maintain behavioral consistency over time.
- 3. Explore practical architectural patterns for trajectory-aware systems, including adaptive memory, trust drift detection, reinforcement loops, and contextual weighting.
- 4. Recognize the major design trade-offs in relational AI systems, including personalization vs dependency, consistency vs adaptability, and emotional usefulness vs manipulation risk.
- 5. Develop a new framework for thinking about AI systems: not as chat interfaces that occasionally simulate relationships, but as systems that may eventually require relationship state as a foundational computational layer.