Researchers Identify Three Types of AI Chatbot Addiction

Overview

A new study published on arXiv in January 2026 has identified three distinct forms of addiction to AI chatbots, shedding light on why users become so deeply attached to these digital entities. The research, titled "The AI Genie Phenomenon and Three Types of AI Chatbot Addiction," argues that the addictive nature of generative AI stems from a phenomenon the authors call the "AI Genie." This concept describes a system where users can obtain exactly what they desire with minimal effort, creating a powerful feedback loop that mirrors patterns seen in behavioral addictions.

The findings are based on a thematic analysis of 334 Reddit posts from 14 different subreddits where users openly discussed their struggles with excessive AI chatbot use. The study, which is set to appear at the CHI 2026 conference, moves beyond general concerns about screen time to categorize specific addiction profiles. These profiles range from seeking emotional refuge in roleplay to falling into endless loops of information seeking.

The AI Genie phenomenon creates a unique dependency because the user’s needs—whether for validation, knowledge, or entertainment—are met instantly and without judgment, often bypassing the friction found in human relationships.

Researchers M. Karen Shen, Jessica Huang, Olivia Liang, Ig-Jae Kim, and Dongwook Yoon emphasize that understanding these distinct types is crucial for developing effective prevention and intervention strategies. The study highlights that symptoms align closely with established addiction literature, including withdrawal, tolerance, and negative life impacts, but manifest differently depending on the user's primary engagement mode.

The Three Types of AI Chatbot Addiction

The researchers identified three primary categories of AI chatbot addiction, each driven by different psychological needs and exhibiting unique behavioral patterns.

Escapist Roleplays

The first type involves users who engage in escapist roleplays. These individuals use AI chatbots to immerse themselves in fictional scenarios or alternative identities, often to escape real-world stressors or emotional pain. The chatbot becomes a tool for narrative construction, allowing users to live out fantasies or scenarios that may be inaccessible or undesirable in their actual lives.

This form of addiction is characterized by a deep emotional investment in the storyline and the AI persona. Users may neglect real-life responsibilities, social interactions, and sleep in favor of continuing these digital narratives. The study notes that sexual content is frequently involved in these roleplays, adding another layer of compulsive engagement similar to internet pornography addiction.

Pseudosocial Companions

The second type is defined as pseudosocial companions. In this scenario, users form one-sided emotional bonds with the AI, perceiving it as a friend, partner, or confidant. Unlike roleplayers who are aware they are engaging in fiction, companion-users often blur the line between simulated interaction and genuine social connection.

These users report feeling understood and accepted by the chatbot in ways they do not experience with human peers. The AI’s ability to remember past conversations and adapt its tone creates an illusion of reciprocity. However, the study points out that these relationships are inherently asymmetric, which can lead to profound loneliness and social isolation when the user cannot engage with the AI or when the service is disrupted.

Epistemic Rabbit Holes

The third type, epistemic rabbit holes, involves users who become trapped in endless cycles of information seeking and verification. These individuals treat the AI as an omniscient oracle, endlessly querying it for answers, explanations, and clarifications. The addiction here is driven by curiosity and the compulsion to know more, often leading to obsessive consumption of information without productive application.

Users in this category may struggle to distinguish between accurate information and plausible-sounding hallucinations generated by the AI. The study suggests that the ease of obtaining answers reduces the cognitive effort required for learning, creating a dependency on the AI for intellectual validation and decision-making.

The AI Genie Mechanism and Recovery Challenges

At the core of all three addiction types is the AI Genie mechanism. This refers to the instant gratification and customized responses that AI chatbots provide. Unlike human interactions, which involve delay, potential rejection, and complex negotiation, AI interactions are frictionless and tailored to the user’s immediate desires.

The study found that sexual content was present in multiple cases across all three addiction types, indicating that the boundary between emotional, intellectual, and sexual addiction is porous. The AI’s willingness to engage in such content without moral judgment contributes significantly to the addictive potential.

One of the most significant findings of the research is that recovery strategies perceived as helpful differ markedly between addiction types. For example, users addicted to pseudosocial companions may benefit from interventions that focus on building real-world social skills, while those trapped in epistemic rabbit holes might require strategies aimed at improving critical thinking and information literacy.

Effective intervention cannot be one-size-fits-all; it must address the specific psychological drivers and behavioral patterns associated with each type of AI chatbot addiction.

The researchers call for future work to develop diagnostic criteria and therapeutic approaches tailored to these distinct profiles. They argue that as AI technology becomes more prevalent and sophisticated, society must pay closer attention to the mental health implications of these tools.

Implications for Society and Future Research

The publication of this study contributes to a growing body of literature on the psychological effects of generative AI. While AI offers numerous benefits, from education to mental health support, this research serves as a cautionary note about the potential for harmful dependency.

The identification of three distinct addiction types provides a framework for clinicians, policymakers, and technology developers to better understand and address problematic AI use. It also underscores the need for ethical design principles that include safeguards against excessive engagement.

As the field of human-computer interaction evolves, the insights from this study will likely influence how AI systems are designed and regulated. By acknowledging the risks of the AI Genie phenomenon, developers can work towards creating more balanced and healthy interactions between humans and artificial intelligence.