Personal Optimization Addiction: When AI Turns Wellness into Anxiety

We connect smartwatches to monitor sleep, heart, and stress, convinced we are improving our health. But what happens when tracking becomes a compulsive obsessio

Until a few decades ago, well-being was perceived as a qualitative state, an intimate and nuanced balance between physical health and mental serenity. Today, we have transformed it into an engineering problem. We strap smartwatches to our wrists, slip biometric rings onto our fingers, and entrust algorithms with the task of monitoring our heart rate variability, oxygen saturation, caloric balance, and REM sleep duration. It is the era of the Quantified Self, a movement born with the intent of increasing personal awareness but which, amplified by Artificial Intelligence, is revealing a disturbing dark side.

The promise of hyper-monitoring is seductive: knowing oneself through data to live better and longer. However, scientific and psychiatric literature is bringing to light a devastating paradox. AI does not generate anxiety in itself; it triggers it the moment it transforms our well-being into a project of infinite and ruthless optimization, in which every vital parameter can always be improved, every deviation from the average becomes a failure, and the body ceases to be lived in and becomes a dashboard to be hacked.

In this in-depth piece for the Scenarios and Reflections column, we will dissect the psychological impact of addiction to personal optimization. Through the most recent clinical reviews, we will explore how the thin line between self-awareness and obsessive-compulsive disorder is blurring, giving rise to new digital pathologies and raising profound doubts about the role of the algorithm in managing our mental health.

1. The Double-Edged Sword of Self-Tracking: From Self-Efficacy to Compulsion

To analyze the phenomenon with intellectual honesty, we must acknowledge that personal monitoring is not born as a pathology. When used with intentionality, it offers measurable benefits. A clinical study conducted on doctoral students and published in PMC demonstrated a positive effect of personal quantification on generalized anxiety; in this context, tracking one's habits and progress increased the subjects' sense of self-efficacy, restoring to them a feeling of control in a highly stressful environment.

The short circuit occurs when tracking transforms from a descriptive tool into a categorical imperative. A qualitative study published in Taylor & Francis ("Too Much of a Good Thing?") captures exactly this boundary. Researchers documented how the obsession with self-monitoring technology coexists insidiously with perceived benefits. Study participants reported the onset of compulsive behaviors: repeated checks dozens of times a day, continuous and obsessive logging of meals, and a serious fixation on numbers (daily steps, calories burned, hours of deep sleep). When the parameters recorded by AI do not reach the pre-set algorithmic goals, users experience acute feelings of guilt, stress, inadequacy, and psychological pressure. Well-being gives way to life performance anxiety.

2. Orthosomnia: Losing Sleep to Measure It

The field in which this dynamic manifests most ferociously is that of nighttime rest. Sleep medicine has recently had to coin a neologism to describe a new clinical condition induced by trackers: orthosomnia.

Derived from the Greek orthos (correct) and somnia (sleep), the term defines a disorder similar to insomnia, but driven exclusively by the obsessive pursuit of optimal sleep data. Sleep tracking algorithms assign a score (the sleep score) to the quality of the user's night. Subjects affected by orthosomnia develop anticipatory anxiety before going to bed, terrified by the idea of "failing" the algorithmic metric. If they wake up during the night, their first thought is not to relax, but to check the clock to calculate how much their score will suffer. The paradox is tragic: the fixation on sleeping perfectly, induced by Artificial Intelligence graphs, causes hyper-activation of the sympathetic nervous system that physically prevents deep sleep.

In these cases, the machine that was supposed to cure insomnia becomes its primary etiological agent.

3. Artificial Intelligence and the Medicalization of the Normal

The integration of increasingly precise biometric sensors with Artificial Intelligence models is shifting the Quantified Self toward permanent clinical surveillance. A systematic review published in PMC ("Wearable Artificial Intelligence for Anxiety and Depression") analyzed dozens of studies, confirming that wearable AI can achieve accuracies above 72% in screening for depressive or anxious states. Even more aggressive systems, such as those based on cross-analysis of personal diaries, chats, and social media presented in an IEEE paper, boast accuracies of up to 92% in predicting mood swings.

Furthermore, research published in Nature Mental Health highlights the promise of hyper-personalized algorithms capable of detecting mental health symptoms in daily life with unprecedented scalability. Multi-agent models, such as those described in Springer, no longer limit themselves to showing heart rate, but cross-reference respiratory rate and HRV to generate true automated clinical reports on chronic stress in real time.

This predictive power, however, opens the doors to an enormous risk, raised by ethical reviews such as "Artificial Intelligence for Mental Health Monitoring: A Solution or a Double-Edged Sword?". Continuous monitoring risks medicalizing normality. Human existence is made of natural physiological fluctuations: an afternoon of sadness, a peak of physiological stress before a meeting, or a restless night are healthy and normal responses of the organism. But if an algorithm isolates them, highlights them in red on a screen, and sends a push notification warning the user of an "anomalous stress peak," that passing emotion is instantly codified as a pre-pathology to be corrected. The sensor does not prevent anxiety: it invents it, providing a diagnosis for a problem the user did not know they had.

4. Therapeutic Chatbots: The Thin Line Between Support and Detachment

In response to the anxiety generated by hyper-modernity (and often by the devices themselves), the technology industry proposes "poison as an antidote": chatbots based on Generative Artificial Intelligence trained to provide psychological support.

Even in this domain, the data are conflicting and fascinating. A rapid review published in Frontiers in Psychiatry showed that chatbots based on the principles of Cognitive Behavioral Therapy (CBT) can effectively produce significant reductions in anxiety and depression symptoms, especially in university students, thanks to 24/7 availability and the absence of human judgment.

However, the meta-analysis published in JMIR scales back techno-utopian expectations. Examining 12 randomized controlled trials (RCTs), researchers found a positive effect, but of a "small-to-moderate" magnitude. The most relevant finding of this study concerns personalization: chatbots with human assistance (hybrid models) and those not excessively personalized perform better than fully autonomous systems. This reminds us of a fundamental principle: greater algorithmic personalization does not automatically translate into greater clinical benefit. Psychological support requires friction, otherness, and sometimes the courage to tell the user an uncomfortable truth—qualities that generative models, trained to please and maximize engagement, struggle to replicate safely.

Key Operational Takeaways (for Users and Developers)

  • Practice Digital Disarmament (For Users): Reintroduce "friction" into monitoring. Disable real-time notifications for stress and sleep metrics. Consult biometric data only once a week to analyze macro-trends, instead of obsessing over the single daily score. Remember that your body knows it is tired even before a titanium ring confirms it.
  • Design for Tolerance (For AI Designers): Health application interfaces must abandon the semantics of failure (red colors, unclosed rings, alert messages for slight fluctuations). AI must be programmed to normalize physiological deviations, communicating to the user that a sedentary day or a sleepless night are part of human nature and do not require immediate "correction."
  • Avoid Algorithmic Determinism: Reject the idea that a quantitative datum is superior to a qualitative emotion. If you wake up feeling rested and energetic, but the algorithm assigns you a sleep score of 45%, believe your body, not the machine.

The Insatiable Algorithm

The push toward personal optimization rests on a profound philosophical misunderstanding: the idea that the human being is an imperfect machine that requires continuous debugging, and that Artificial Intelligence is the source code to achieve a fabled biological perfection.

AI makes continuous monitoring invisible, surgical, and relentlessly predictive. It can undoubtedly anticipate cardiac crises or acute depressive states, saving human lives. But when applied indiscriminately to daily life, it transforms it into an infinite series of metrics to be corrected, generating a sense of chronic inadequacy. The machine does not know the concept of "enough": for a maximization algorithm, there will always be one more step to take, a more regular heartbeat to achieve, one less gram of fat to burn.

Faced with this gilded cage made of high-resolution graphs and biometric notifications, the question we must ask ourselves is decisive: if my sense of self-esteem, my mood, and my serenity depend on the score an algorithm decides to assign to my day, am I truly taking care of myself, or am I tragically delegating my mental health to a mathematical system that will never be able to tell me "you're fine just as you are"?

Bibliographic References and Sources

Article by the Editorial Team of La Bussola dell'IA