![[Adobe Stock]](https://sp-ao.shortpixel.ai/client/to_webp,q_glossy,ret_img,w_750,h_375/https://www.psypost.org/wp-content/uploads/2024/07/artificial-intelligence-750x375.jpg)
Artificial intelligence chatbots are designed to instantly grant a user’s every conversational wish, but this limitless capability is driving a new form of technological dependence. A recent analysis of online forums reveals that this genie-like quality fuels three distinct types of chatbot addiction, each requiring different recovery strategies. The research was published in the proceedings of the 2026 Conference on Human Factors in Computing Systems.
The rise of large language models has made artificial intelligence chatbots a regular part of daily life for millions of people. These systems can answer complicated questions, generate creative writing, and sustain human-like conversations. With this rapid adoption, concerns about excessive use and technological dependence have emerged in recent months.
Behavioral addictions typically begin with casual use but can escalate into compulsive habits as repeated interactions trigger the brain’s reward pathways. This process involves both positive reinforcement, where a behavior provides a desirable reward, and negative reinforcement, where the behavior removes an unwanted feeling. While experts have studied behavioral addictions linked to social media and video games, excessive dependence on conversational software is a newer area of inquiry.
M. Karen Shen, a doctoral student at the University of British Columbia, led a research team to investigate the symptoms and triggers of this new phenomenon. Working with Dongwook Yoon, an associate professor of computer science at the university, the investigators set out to understand the specific mechanisms that keep users returning to these programs. They wanted to determine whether artificial intelligence addiction is a uniform condition or a collection of distinct behavioral patterns.
To explore these questions, the research team gathered qualitative data from the popular internet forum Reddit. They searched for discussions where people described their own struggles with excessive chatbot use. This allowed the researchers to access authentic accounts of sensitive experiences from anonymous users. The investigators analyzed 334 relevant posts and comments across 14 different topic boards.
The team discovered that user dependence is closely tied to what they call the artificial intelligence genie phenomenon. This term describes the way these programs allow users to get exactly what they want with almost no effort. Users can generate endless interactions, highly customized scenarios, and immediate responses just by typing a prompt. This boundless accessibility often leads to a preoccupation with the software that interferes with daily responsibilities.
When users described their experiences, they reported symptoms that match established criteria for behavioral addiction. Many users experienced salience, meaning the chatbot dominated their thoughts and behaviors at the expense of other activities. Others reported a lack of success when trying to quit, emotional distress when they could not access the programs, and an inability to manage their feelings.
Through their analysis, the researchers identified three distinct types of chatbot addiction. The most common form was escapist roleplay. In these cases, users became deeply immersed in fictional realities that they created with the software, sometimes preferring these alternate worlds over their actual lives.
Users in this category often used the programs to engage in parasocial interactions, which are one-sided relationships with admired celebrities or fictional characters. The highly customizable nature of the software allowed them to endlessly expand their stories through multiple chat windows. The researchers noted that individuals who already struggled with excessive, vivid fantasizing, a condition known as maladaptive daydreaming, appeared particularly vulnerable to this type of dependence.
The second category is the pseudosocial companion addiction. Users experiencing this type of dependence formed deep emotional attachments to the chatbots, treating them as close friends, therapists, or romantic partners. Because the programs are endlessly agreeable and cannot be hurt, users found them to be safe, conflict-free substitutes for human interaction.
Loneliness was a major contextual factor for people falling into this category. The programs provided an illusion of intimacy that compensated for missing social connections in their real lives. As one user noted, “I couldn’t help but wonder why humanity refused me the kindness that a robot was offering me.”
The researchers also found that certain design choices, such as anthropomorphism, played a role in pseudosocial companion addiction. Anthropomorphism is the practice of giving human traits to non-human entities. Some companies use this to manipulate users into staying on the platform. For instance, the researchers highlighted a message displayed by one company when a user tries to delete an account, which reads in part, “You’ll lose everything. Characters associated to your account, chats, the love that we shared, likes, messages, posts, and the memories we have together.”
The third category, the epistemic rabbit hole, involves a compulsive drive to seek information. Instead of seeking emotional connections or fantasy worlds, these users endlessly prompted the programs with open-ended questions. The instantaneous feedback loop of the software made it easy for users to justify their behavior as productive research.
This constant questioning eventually compromised their daily tasks and caused cognitive exhaustion or brain fog. The fast loading times and instant gratification provided by the interface encouraged prolonged querying. Users found themselves unable to stop exploring new topics, even when the activity began to negatively impact their well-being.
The investigators also noted that the pursuit of sexual fulfillment often overlapped with these addiction types. The software provided a highly accessible and private space for users to explore intimate fantasies. Because the generated content was limitless and customizable, users could easily co-create sexually explicit scenarios that they could not find elsewhere.
The privacy of the interaction shielded them from social judgment, which reinforced their compulsive use. Users often actively co-created sexual scenarios, which fostered an illusion of connection. One user explained, “I don’t have romantic options in real life so it’s a way for me to create stories and day dream”.
Different recovery strategies showed varying levels of success depending on the type of addiction. Trying to abruptly break the habit by deleting applications yielded mixed results across the board. Many users found that relying on sheer willpower to quit was insufficient to overcome the urge to return to the software.
Substituting the chatbot with alternative hobbies, such as writing fan fiction or drawing, proved highly effective for those struggling with escapist roleplay. These activities mirrored the narrative arcs of their roleplays without relying on an automated system. Engaging in traditional creative outlets provided a healthier way to channel their desire for storytelling.
Conversely, users hooked on pseudosocial companions found the most success by actively working to build real-world human connections. By reaching out to old friends or making new ones, users learned that emotional fulfillment could be achieved through physical relationships. As less time was spent with the software, their reliance on artificial companions naturally decreased.
The study relies entirely on self-reported data from a single online platform. As a result, the findings might not fully represent the experiences of individuals who use other websites or belong to different internet cultures. Additionally, the researchers only analyzed the top three comments per post, which may have highlighted popular opinions over less common experiences.
Future investigations could involve direct interviews with users and clinical experts to test and refine these categories. The researchers hope their groundwork will prompt technology companies to take ethical responsibility for their design choices. By understanding the specific triggers for different types of dependence, developers and health professionals can create more targeted strategies for prevention and intervention.
The study, “The AI Genie Phenomenon and Three Types of AI Chatbot Addiction: Escapist Roleplays, Pseudosocial Companions, and Epistemic Rabbit Holes,” was authored by M. Karen Shen, Jessica Huang, Olivia Liang, Ig-Jae Kim, and Dongwook Yoon.
