Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
As artificial intelligence becomes increasingly persistent, personalized, socially situated, and eventually embodied, a centralproblem for human-AI interaction is shifting from whether people can respond relationally to computational systemstoward what allows those relationships to remain recognizable across time. Recent research on AI-companion disruption,platform migration, model replacement, feature withdrawal, and service termination suggests that technical persistence,memory persistence, perceived identity continuity, and relational continuity are not equivalent. Conversation archivesmay survive while familiar interaction disappears, and a system may retain or improve functional capabilities whilechanges in memory behavior, interactional style, relational stance, or access produce substantial perceived rupture (Banks,2024; De Freitas et al., 2026; Do et al., 2026; Yuan et al., 2026). This paper develops the AREN Recognition Continuity Hypothesis: Human-AI relational meaning may emergethrough the repeated experience of being recognized by a sufficiently continuous counterpart across time, provided thataccumulated history remains revisable by the continuing person. The hypothesis develops from a wider Aeon Mundi andAREN trajectory concerning relevant asymmetry, operative history, relational architecture, irreducibility, historicaldefeasibility, mutual legibility, and reciprocal maintenance (Lambdin, 2026a-f). The proposed model distinguishesinfrastructural continuity, autobiographical continuity, interactional continuity, relational-stance continuity, andsituated recognition, while treating perceived identity continuity not as another technical subsystem but as an emergenthuman judgment concerning whether the counterpart encountered now remains meaningfully connected to thecounterpart encountered before. The paper further argues that continuity alone is insufficient. A system that remembers extensively may become lessrelationally intelligent if accumulated information about a person hardens into a predictive representation that resistscorrection by the person themselves. Evidence from multimodal affect recognition reinforces this limitation bydemonstrating that observable expression and inferred affect can diverge from self-reported experience, underscoring theimportance of epistemic humility in systems attempting to recognize human states (Dragut et al., 2026). Relationalcontinuity must therefore preserve both history and defeasibility. This tension yields the design principle around which thepaper converges: Remember enough to recognize; remain uncertain enough to rediscover. KEYWORDS relational continuity; relational intelligence; human-AI relationships; AI companions; situated recognition; historicaldefeasibility; memory; identity continuity; embodied AI; companion robotics; affective computing; human-robot interaction; AREN; AeonMundi
No comments yet — start the discussion below.