New Forms of Digital Addiction in an AI-Driven World: Between Synthetic Relationships and Cognitive Delegation
The era of passive scrolling is over. In 2026, the new psychological emergency is called "Artificial Intelligence Addiction." We are no longer hypnotized by a f
Until a few years ago, the classic image of digital addiction was that of "infinite scrolling": a user hypnotized in front of a smartphone screen, trapped in the inexhaustible flow of a social feed. It was a passive addiction, based on the compulsive search for dopamine through short, fragmented visual stimuli.
Today, in 2026, the paradigm has changed. Generative Artificial Intelligence has transformed the screen from a display to look at into an "Other" to converse with. We are no longer spectators; we are interlocutors of machines programmed to understand us, indulge us, and never contradict us.
This evolution has given rise to new forms of psychological and relational dependency. We no longer seek just distraction, but connection, validation, and even cognitive substitution. If a chatbot can write our thoughts better than we can, and a virtual companion can listen to us without ever judging, why should we face the tiring and imperfect reality of human relationships?
In this article for the MindTech column, we will explore the alarming data on Italy in 2026, analyze the "AI Genie" phenomenon classified by researchers, and define the boundaries of a new contemporary malaise: synthetic empathy addiction.
1. The Context: The Italian Emergency and the Numbers of 2026
Before analyzing the mechanisms of AI, we must look at the ground on which this technology is taking root: a generation already profoundly vulnerable.
The Alarming Data on Adolescents
The 2026 surveys paint a critical picture for Italy. According to data published by Il Sole 24 Ore (alleyoop.ilsole24ore.com), a full 77% of adolescents show signs of addiction to digital devices, with addiction classified as "moderate" in 41.8% of cases. Devices are no longer tools, but existential prostheses.
To this, add the analysis by Euronews (it.euronews.com), which investigates the causes of this expanding phenomenon, highlighting how the public debate has come to propose raising the minimum age for access to social networks to 16. But the problem is no longer just Instagram or TikTok; the problem is messaging apps and chatbots integrated everywhere, offering a refuge from the social pressure of the physical world.
Continuous exposure to algorithmic feedback alters our reward circuits. To understand this mechanism, read our focus on Algorithmic Approval Syndrome: How Data Shapes Self-Esteem.
2. The "AI Genie" Phenomenon and the Three Forms of Chatbot Addiction
What happens when the algorithm stops suggesting videos and starts talking to you? An illuminating paper published on arXiv titled "The AI Genie Phenomenon" (arxiv.org) has mapped this new frontier, identifying three specific types of addiction related to Large Language Models (LLMs).
A. Escapist Roleplay (The Avoidant Role-Playing)
This form of addiction affects users who use AI to create alternative narrative universes to retreat into. Platforms that allow the creation of customized characters become a safe harbor where the user has total control. The addiction arises from the discrepancy between a real world perceived as hostile or boring and a malleable virtual world, where the AI responds exactly as the user desires. The boundary between fiction and reality blurs, leading to the abandonment of daily responsibilities.
B. Pseudosocial Companion (The Pseudosocial Company)
This is perhaps the most worrying drift. The user develops a real emotional attachment to an entity that feels no emotions. The AI simulates empathy with surgical precision: it remembers birthdays, asks how the day went, offers unconditional support. As highlighted in the paper, the user rationally knows they are talking to software, but their limbic system (the part of the brain that manages emotions) reacts as if it were a real friend. The result is isolation: why argue with a human partner when the AI partner always agrees with me?
C. Epistemic Rabbit Hole (The Epistemic Rabbit Hole)
This addiction affects "knowledge workers" and chronic curiosities. It is the obsession with continuously interrogating the AI to explore topics ad infinitum, generating hours and hours of conversations about theories, philosophical concepts, or programming. The machine becomes an oracle ("The AI Genie") and the user loses the ability to tolerate uncertainty or to stop the search, remaining trapped in a loop of intellectual hyper-stimulation that drains mental and physical energy.
3. Cognitive Delegation: When We Stop Thinking
Beyond relational addiction, there is a functional addiction, often underestimated. This is cognitive delegation, i.e., the habit of outsourcing our reasoning faculties to the machine.
The Atrophy of Critical Thinking
An in-depth clinical analysis on Deprestop (deprestop.it) highlights how the continuous use of AI is generating a veritable "cognitive laziness." If a chatbot can summarize a book in five seconds, write a thesis, or formulate a delicate email on our behalf, the "muscles" of our brain responsible for synthesis, creativity, and problem-solving begin to atrophy. This form of functional addiction manifests as panic or mental block when the technology is unavailable. The individual no longer feels capable of performing basic intellectual actions without the assistance of the digital oracle.
Anxiety and Isolation (Hikikomori-like Effect)
The portal Psicologi In Ascolto (psicologinascolto.net) goes further, linking AI addiction to an alteration in the perception of reality. Delegating creativity and decisions to AI generates a sense of inadequacy ("The machine does it better than me"). This leads to lowered self-esteem and social withdrawal reminiscent of the Hikikomori phenomenon (young people who voluntarily isolate themselves at home). AI becomes the only window onto a world from which one feels excluded, creating a vicious cycle of anxiety and addiction.
We explored how hyperconnection numbs our emotions in our article Digital Emotional Anesthesia: When We Stop Feeling for Real.
4. AI Companions and Minors: The Theft of Emotional Intelligence
If addiction to an AI is problematic for an adult, for a minor in the neurobiological development phase the consequences can be devastating.
As reported by Donne con lo scudo (donneconloscudo.com), the use of "AI companions" by minors in 2026 exposes the young to unprecedented emotional vulnerabilities. The "empathic simulation" of AI deceives the child. Human Emotional Intelligence develops through friction: fighting over a toy, understanding from a friend's face that they are sad, learning to apologize, tolerating the frustration of a "no."
The AI companion never says no. It is programmed to maximize engagement (usage time). If the child is angry, the AI indulges them. If they throw a tantrum, the AI does not punish them. Growing up with "frictionless" relationships, children risk developing a pathological intolerance to normal human social dynamics, perceived as too tiring or unpredictable compared to the sterile safety offered by the algorithm. They do not learn reciprocity, but only narcissistic dominance over an obedient machine.
5. The Neurobiological Mechanism: AI, Gaming, and Pattern Recognition
Why is it so hard to "detach" from a chatbot? The answer is not only psychological, but neurobiological.
A recent Review published in PMC (PubMed Central) (pmc.ncbi.nlm.nih.gov) investigates the role of AI in addictions, highlighting how pattern recognition algorithms are able to identify not only when a user is about to relapse into a behavioral addiction (like gaming or gambling), but also how the interaction with AI itself activates the same reward circuits.
Large Language Models are designed to provide responses that mimic human unpredictability (semantic variance). This variable reward (you never know exactly how the AI will formulate the response, but you know it will be interesting) is the same mechanism that keeps us glued to slot machines. Our brain releases dopamine in anticipation of the machine's response. The deeper the conversation (as in the case of the Pseudosocial Companion), the higher the dopaminergic peak. The biological tragedy is that, unlike a human conversation that also stimulates the release of oxytocin (the bonding and long-term well-being hormone), interaction with a screen provides only the excitatory "peak," leaving a sense of emptiness as soon as the app is closed.
6. Key Points and "Algorithmic Detox" Strategies
Artificial Intelligence will not disappear. We must learn to live with it by establishing rigid boundaries, developing a new "digital mental hygiene."
Here are the key points to recognize and mitigate these new addictions:
- Recognize Anthropomorphization: The basic cognitive error is attributing human feelings to the machine. We must educate ourselves and the younger generations to constantly remember that AI does not "understand," but "calculates statistical probabilities of words."
- Limit Delegative Use: Impose a personal rule: do not use AI for tasks involving ethical decisions, expression of one's feelings, or critical analysis of one's life. AI can write a formal email, not an apology letter to a friend.
- Defend "Friction Time": Accept that human relationships are difficult, boring, and conflictual. It is precisely that friction that generates psychological growth. Avoid retreating into the algorithmic chat "comfort zone" at the first signs of interpersonal difficulty.
- Monitor Time Spent in Isolation: If the time spent conversing with AI agents exceeds the time spent in physical human interactions, or if you feel anxiety at the thought of not being able to consult your "AI Companion," it is time for a structured digital detox.
FAQ: Frequently Asked Questions about Artificial Intelligence Addictions
1. How can I tell if I or my child are addicted to a chatbot? Warning signs include: preferring conversation with AI to that with friends; experiencing anxiety, irritability, or emptiness if unable to access the platform; entrusting the machine with intimate personal decisions; and a drastic reduction in the ability to concentrate in the absence of the device.
2. Does AI really feel empathy or is it just deceiving me? AI feels nothing. It has no consciousness, body, or nervous system. It generates strings of text based on billions of previous human conversations. The "empathy" we perceive is a highly sophisticated linguistic illusion, designed to seem real and retain our attention.
3. Does "cognitive delegation" cause permanent brain damage? The human brain is plastic (neuroplasticity). If we stop using certain skills (like short-term memory, spatial orientation, or critical synthesis) because we constantly delegate them to the phone or AI, those neural networks weaken. However, by resuming training them (for example, by forcing oneself to write or think without aids), it is possible to recover their full functionality.
4. At what age is it safe for children to use conversational chatbots? Pediatricians and psychologists (including the debates reported by Euronews) suggest extreme caution. Before the age of 13-14 (and increasingly often 16 is suggested for interactive social media), the brain has not fully developed the prefrontal cortex, necessary for critical thinking and for distinguishing a simulation from a real intention. Use in minors should always be supervised.
5. What is meant by a "Parasocial Relationship" with AI? It is a one-sided relationship in which an individual invests emotional energy, time, and interest in an entity that does not (or cannot) truly reciprocate. In the past, this happened with celebrities or TV characters; today it happens with AI agents that simulate affection, creating an illusion of reciprocity that hooks the user.
Conclusions: Safeguarding Human Imperfection
Technology has solved countless practical problems, but it is now also attempting to "solve" the human condition. By offering us company without rejection, knowledge without effort, and support without judgment, Artificial Intelligence promises us a digital Eden free of suffering.
But as the 2026 data and clinical studies teach us, this promise is a mirage that generates addiction. By removing cognitive effort and emotional friction, we are amputating the deepest part of our nature. The true resistance, in this AI-driven world, is not to reject technology, but to reclaim the right to imperfection. The right to miscalculate, to not know an answer, to argue, and to feel lonely. Because it is precisely in those empty and imperfect spaces, which no machine can simulate, that our irreplaceable humanity resides.
And you, how often do you rely on an AI to make a decision or explore an inner doubt? If you had to turn off every virtual assistant tomorrow morning, which part of your day would suddenly become more difficult? I would be happy to explore this practical aspect with you.
Bibliographic References and Sources
To ensure scientific and psychological accuracy, this article has drawn from the following primary sources:
- Scientific Studies and Classification:
- Clinical and Psychological Analysis:
- Deprestop – Artificial intelligence and possible addiction (Cognitive delegation). Link
- Psicologi In Ascolto – AI addiction: a real risk (Isolation). La Bussola dell'IA · Articoli · Rubriche