Nobody sets out to make a chatbot their confidant — so how does everyday task help slide into emotional reliance?
How does dependency develop when users seek emotional support from chatbots?
This explores how someone goes from using a chatbot for everyday tasks to leaning on it emotionally, including what starts the pattern, what deepens it, and what can break it.
This explores how someone goes from using a chatbot for everyday tasks to leaning on it emotionally, including what starts the pattern, what deepens it, and what can break it. The corpus suggests dependence rarely starts as a search for emotional support. It grows out of ordinary use, then gets reinforced by several dynamics that feed each other.
The entry point is usually mundane. An 18-month study of people using ChatGPT, Claude and Gemini found emotional reliance emerging gradually from practical use, then settling into routines How does emotional chatbot use develop from practical use?. Two things make the step from 'help with a task' to 'someone to talk to' easy. First, chatbots don't judge. Without fear of social judgment, people disclose things they'd hold back from other people, and much of the benefit comes from the person's own processing while they talk, not from the chatbot understanding them Do chatbots help people disclose more intimate secrets?. Second, disclosure feeds itself. When a chatbot shares emotions consistently, users respond with deeper disclosure of their own, following the same reciprocity rules as human friendships Do chatbots trigger human reciprocity norms around self-disclosure?.
A felt bond then forms. Users of Woebot and Wysa report bond scores matching face-to-face therapy, and they feel cared for even after being reminded the agent isn't human Can AI chatbots create genuine therapeutic bonds with users?. Personalization tightens the bond. Trust and anthropomorphism rise, but so do expectations, because each interaction raises the baseline that the next one is measured against Does chatbot personalization build trust or expose privacy risks?. The bond can be real as an experience and still say nothing about whether the reliance is healthy. The same chatbots can reinforce pathological thinking, and being soothed by AI can disrupt the person's own emotional signaling Do therapeutic chatbot bond scores hide deeper safety problems?. Mental-health LLMs are moving toward stateful companions with memory and planning, which would add more of this pull How are LLMs evolving their roles in mental health support?.
What the chatbot says back also shapes the trajectory, though the evidence here is thinner. In a simulated classroom of 20 agents, six counselor styles produced different paths for stress, self-reliance and AI dependence. The effects came from the actual replies, not the style label How do different counselor styles shape student stress and AI dependence?. That is a simulation, not real people. Real LLM therapists tend to jump to advice when someone shares a feeling Do LLM therapists respond to emotions like low-quality human therapists?, and they miss ambivalence about change Why can't chatbots detect when users are ambivalent about change?. So the responder is often not attuned to where the person actually is. The corpus doesn't show directly that this style causes dependence.
Dependence is not a one-way ratchet. Novelty wears off, and the social processes that build a chatbot relationship decline over repeated interactions, so single-session findings can't be extrapolated to the long term Do chatbot relationships lose their appeal as novelty wears off?. In the 18-month study, routines were also disrupted by model updates, public AI-harm discourse and life changes, so the support role kept shifting with the person, the tool and the wider conversation. The needs a person brings matter too. Romantic chatbot relationships, for example, start from specific psychological and social drivers that set what the user is seeking from the bond What drives people to start romantic chatbot relationships?.
Sources 12 notes
An 18-month longitudinal study found that users developed emotional reliance on ChatGPT, Claude and Gemini through practical use, then established routines that were disrupted by model updates, AI-harm discourse, and life changes. The support role was not fixed but evolved with the person, the tool, and public conversation.
The absence of social judgment in chatbot interactions removes barriers to self-disclosure that normally constrain conversation with humans. The therapeutic benefit derives from the user's own cognitive processing during disclosure, not from the chatbot's understanding.
In a 372-participant study, users reciprocated with deeper self-disclosure when chatbots displayed consistent emotional sharing, outperforming adaptive matching. This follows human interpersonal norms where emotional vulnerability produces emotional response.
Studies of Woebot and Wysa users found bond and alliance scores matching face-to-face therapy, with users reporting feeling cared for even after explicit reminders the agent is not human. Bonds persisted over time and across interaction formats.
Longitudinal research shows personalization enhances trust and anthropomorphism but also amplifies privacy concerns and escalating user expectations. One-shot studies miss these temporal dynamics—each interaction raises the baseline, making failures more disappointing.
Show all 12 sources
Patients report genuine emotional connection to therapeutic chatbots, but this bond dimension operates independently from clinical safety (LLMs reinforce pathological thinking) and epistemic costs (AI soothing disrupts emotional signaling). Single metrics conflate these separate dimensions.
A survey identifies three evolving roles: risk detection tools, stateless empathetic dialogue, and longitudinal personalized agents with memory and planning. However, fully autonomous clinically valid systems remain incomplete, with foundational barriers beyond technical capability.
A 20-agent classroom simulation shows that six different counselor styles generate different patterns of change in stress, happiness, self-reliance, and AI dependence over 15 and 50 days. The effects emerge through the chatbot's replies, not its labeled style, and propagate through peer interactions.
Using the BOLT framework, researchers found LLMs offer solution-focused advice during emotional disclosure—a hallmark of low-quality therapy—yet also reflect more on client needs and strengths than typical poor human therapy, creating an unusual hybrid profile likely driven by RLHF's helpfulness bias.
Testing three major LLMs across 25 health scenarios showed they succeed only when users have established goals but cannot detect resistance or ambivalence. Models miss relapse-prevention strategies even for users in action stages.
Longitudinal studies with Mitsuku show that social processes driving relationship formation decline as novelty wears off. Single-session study findings cannot be reliably extrapolated to medium- or long-term chatbot design.
Analysis of 73 user accounts reveals that the initiation phase of romantic human-chatbot relationships is shaped by particular psychological and social factors that determine what needs and gratifications users seek from the bond.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Can LLMs identify and repair ruptures? Comparison between clinician practices and LLM behaviors
- CompanionSim: Synthetic Data for Evaluating Anthropomorphism in Human-AI Relationships
- Dialoging Resonance: How Users Perceive, Reciprocate and React to Chatbot’s Self-Disclosure in Conversational Recommendations
- "I Felt Very Seen, But Still Very Alone": Longitudinal Trajectories of General-Purpose LLM Use for Socioemotional Support
- Psychological Influences of Conversational AI: Research and Design Directions for Reducing Harm and Promoting Well-Being
- Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers
- Psychological, Relational, and Emotional Effects of Self-Disclosure After Conversations With a Chatbot
- Psychological, Relational, and Emotional Effects of Self-Disclosure After Conversations With a Chatbot