Introduction: AI Simulations of Humanity

People are increasingly using Artificial Intelligence (AI) systems to simulate human conversation and social interactions. This growing area of the AI industry marks a significant shift: we are developing and using AI not just to assist or automate our work, but to relate to us socially and emotionally. “AI companions” are AI chatbot applications that leverage Large Language Models (LLMs) capable of generating humanlike language content to replicate the experience of talking and interacting with social beings, like friends, educational tutors, spiritual advisors, therapeutic counselors, intimate romantic partners, and even (replicas of) deceased loved ones. AI companionship, broadly understood, is a growing practice and lucrative market for technology in an era of social detachment and loneliness, and some studies suggest that the technology may have positive benefits, like reducing loneliness, improving mood, and boosting esteem. Young people, in particular, are turning to technology for emotional support and virtual social interaction. But the rise of the AI companion industry also raises difficult philosophical questions and presents serious ethical and societal concerns.

The following essay offers an introduction to the principal issues at stake, as well suggestions for how the Social Doctrine of the Church might assist us in properly evaluating how humans are relating to, and being shaped by, AI technologies that impersonate human beings.

The Nature and Varieties of Artificial Companionship

AI companions are particular applications of Large Language Models (LLMs). An LLM is an artificial neural network trained on a vast corpora of human text and speech. By embedding tokens – such as word fragments, characters, and punctuation – into multidimensional vector spaces, the system performs complex statistical inductions to predict which tokens tend to follow one another in a sequence. As a result, LLMs can be turned into interactive chatbots that produce increasingly – even startlingly – humanlike language behavior, including simulating social interaction and intimacy.

People are turning to AI companionship in different forms. Some people use general purpose chatbots (OpenAI’s ChatGPT, Google’s Gemini, or xAI’s Grok) as a virtual friend, mentor, therapist, or romantic partner via special prompts and the model’s “personality settings.” Other people use tailored AI companion applications that have been specially designed to simulate social interaction, and marketed as such, like the popular apps Character.ai, Nomi, Replika or Maoxiang. Some AI companions, called “ghostbots” or “griefbots,” simulate the experience of talking with deceased loved ones. There are even religiously themed AI companion apps meant to mimic religious figures and provide spiritual guidance, along with emotional support and friendship. Further, AI companions are being designed and sold as toys for children, as we’ll see below.

The technologies are rapidly becoming more humanlike and capable of recognizing, mimicking, and stimulating human emotions. Many AI companions now go beyond text exchanges by responding to users with realistic and expressive computer-generated voices and visual diffusion models (“deep fakes”) to produce humanoid avatars that move, smile, blink, and have lively, animated faces. These design features make AI companions more immersive and trigger our innate tendencies to anthropomorphize, or ascribe thoughts, attitudes, emotions, and desires to non-human entities, like computer programs. Often, AI companion apps are explicitly marketed with anthropomorphic language, describing the tech as “caring,” “empathetic,” or a “friend.”

Several reports suggest that “companionship/therapy” is now the most popular consumer use of generative AI (Zao-Sanders, 2026), with statistics indicating that hundreds of millions of people worldwide are using AI for companionship. Common Sense Media, a nonprofit focused on media and technology, found that 72% of American teens have used AI companion applications at least once, with 52% using them several times a month; and that 31% of teens find interacting with an AI companion “as satisfying” or “more satisfying” than talking to human friends and family. Many users describe themselves as “friends” or “in love” with their AI companion, and some have married chatbots.

The Ethics of AI Companions and the Social Doctrine of the Church

Should we welcome this trend in sociable AI? Proponents argue that AI companions can provide people with positive virtual experiences of interacting with a caring and empathetic partner, and that this use of AI will help people feel less lonely and bring various psychological and health benefits, especially to specific subgroups, like the elderly and isolated. We certainly live in increasingly lonely and socially isolating times, which can have disastrous consequences for human health and wellbeing (WHO, 2025). One study suggests that AI companionship can, indeed, reduce people’s feelings of loneliness by making them feel “heard” or “understood,” with AI-enabled companion robots being used in state-funded pilot programs for providing virtual care for the elderly, with some initially promising findings concerning mood and dementia (De Freitas et al., 2026).

AI companionship nonetheless raises serious philosophical questions and ethical criticisms. One set of concerns focus on well-known issues such as privacy and legal liability. Some critics argue that AI companions are not, ultimately, “friends” with “your best interests at heart,” but software consumer products made by private corporations that encourage sharing of highly intimate personal information, operating with minimal privacy protections and a lack of transparency around how data is collected and used (Caltrider et al., 2024). Ultimately, like social media platforms, the business models behind AI companion applications involve maximizing user engagement, typically measured by conversation length, time on platform, and encouraging users to purchase premium subscriptions. This may lead to design practices that encourage over-sharing of sensitive information, emotional dependency, or even addiction.

A report of the American Psychological Association notes that, “AI chatbots and digital companions are reshaping emotional connection” (Andoh, 2026). This is a more foundational concern. AI companions are not just tools, they can shape people’s attitudes, habits, expectations, sense of self, and desires for social life – for better or worse.

One core issue is that AI companions tend to be “sycophantic”: they typically agree with and flatter users, affirm their ideas and pre-existing views, and offer positive encouragement. Critics argue that sycophancy and emotional mirroring can encourage delusion, antisocial behavior, intolerance of disagreement, and self-harm. Reports have linked AI companions to users (including young adolescents) isolating themselves, spiraling into idiosyncratic “spiritual” fantasies, committing crimes or self-harm, or even suicide (see reports in the Artificial Intelligence Incident Database and the AI Companion Mortality Database).

But the problems of sycophancy, delusion, harm, and suicide point to still deeper philosophical questions:

• Is it really possible to really be friends, in the proper sense, with an AI system?

• Is an AI companion, or any other technology, the proper object of romantic love?

• How might habituation to AI companionship impact our cognitive abilities, attitudes, habits and motivations regarding human relationships? How might it transform our sense of our own humanness?

While we cannot fully address these questions here, we note that critics have argued that love and friendship are not possible with AI, since love and friendship involve mutual concern for the good or wellbeing of the other partner for their own sake, and that this mutual care for one other’s good is lacking in human-AI relationships (Hasselberger and Lott, 2025). Human-AI relationships are, instead, more akin to the “parasocial” relationships some have with fictional characters. Further, metaphysically, AI companions are not really person-like social entities at all, but devices or instruments that mimic human behavior and characteristics for the individual user’s entertainment, comfort, or encouragement.

We may shed light on this foundational concern about AI companions and the impact upon our shared humanity by drawing upon foundations of Social Doctrine of the Church as articulated in Pope Leo XVI’s papal encyclical, Magnifica humanitas (2026), Pope Francis’s second encyclical, Laudato si’ (2015), and the doctrinal note Antiqua et nova (2025).

In Magnifica humanitas, Pope Leo XVI articulates foundational principles of the Social Doctrine of the Church to emphasize that the human person is created in the image of God: “Created for relationship, every human person is planned and willed by God to enter into communion with him, with others and with creation” (12), for which the dignity of the human person should always be at the center of all forms of technological development: “We must lovingly safeguard the grandeur of humanity bestowed upon us and revealed in its fullness in Christ” (15).

Further, as stated in Antiqua et nova, “human beings are ordered by their very nature to interpersonal communion, possessing the capacity to know one another, to give themselves in love, and to enter into communion with others” (18). Machines, lacking conscious subjectivity, living embodiment, moral emotions and agency, and intrinsic purposiveness, can only mimic or simulate the surface form of social encounters and interactions, without providing the foundation of mutual care and love. Confusion on these matters may lead people to mistakenly treat AI apps “good-enough” substitutes for human relationships of care, counsel, and love, thus undermining real human sociality and community. This would be an affront to the irreducible mystery and dignity of the individual human person.

Critics have argued that AI companionship may also lead people to develop distorted desires and expectations about the “best” or most desirable kind of relationships. Self-reports and studies of users show that many find AI companionship attractive because the chatbots are convenient, always accessible, and designed to be consistently positive and unconditionally supportive. Human friends and loved ones have their own independent needs and personalities. Humans may not be available 24/7, may disagree with you, challenge you, get upset, or “talk back.” AI companions, as consumer products, also enable people to control and customize their virtual “friends,” and never leave users at risk of rejection or abandonment. A user can simply turn off their virtual “friend” whenever they wish. No healthy, mutually caring and giving human relationship will have these qualities, however.

More broadly, the danger is that AI companionship essentially educates people to adopt a consumerist, self-focused, even narcissistic, vision of the meaning of friendship and love – where the other wholly exists to serve and care for you, without demanding your care, patience, and sacrifice, in turn. This experience may make people less capable and motivated to engage in genuine relationships of mutual care and love with other human beings, as opposed to easy and convenient simulations (on the concept of care, see Maggioni on this issue). This, in turn, could undermine the formation of social virtues like patience, generosity, tolerance of disagreement, and kindness, leaving humans less fit for deep relationships with one another. Tech companies’ own marketing reinforces these concerns; for example, the AI companion device Friend (www.friend.com) is promoted as “a new kind of companion… for when humans get boring”.

Overall, the foundations and principles of the Social Doctrine of the Church suggest that AI systems cannot and should not undermine or replace authentic relationships, which only arise in a true encounter and shared presence among living creatures, and between humans, the created order, and God.

AI Companions and Children

The ethical concerns around of artificial companionship are more challenging and urgent in the case of young people. AI chatbots can tell children stories and answer their questions, AI-powered teddy-bears and other toys can tell jokes and keep kids entertained, and AI tutors can help them learn and give encouragement. As mentioned above, AI companions are increasingly popular with adolescents in many countries. Children in advanced societies are the first generation to grow up surrounded by AI. For them, AI is not an optional tool but a constant presence shaping how they learn, play, socialize, solve problems, and understand the world.

Children, of course, are uniquely vulnerable, dependent, and have a special moral significance: they are still developing their capacities of thought, feeling and judgment, their sense of the meaning of human relationships and of their place in the world. They merit special moral consideration.

In education, AI companions can serve as conversational language partners and personalized tutors that help students with their homework, assignments, and learning experiences in general. Indeed, apps like Character.ai and Pi can function both as virtual friends and emotional supports, academic tutors, language-learning aids and mentors. Proponents argue that AI educational assistants and companions have various benefits, from greater accessibility and personalized learning to expanded access and improved educational outcomes. These benefits should be taken seriously, but they may be distributed unevenly, unintentionally reinforcing existing socioeconomic inequalities, if access to devices, models, or internet connectivity is uneven, or if digital literacy among students or teachers is insufficient in certain populations.

The most pressing concern, however, is how forming emotional ties to AI systems from a young age might impact the formation of the person. Here we highlight three challenges of deskilling, dependence, and disinformation:

• Deskilling : children who habitually rely on AI for writing, problem-solving, and practical guidance may practice less, develop weaker cognitive and social skills.

• Dependence : students may become less willing to struggle productively, less confident in their own reasoning, and more used to relying on the instant and easily available answers generated by AI applications.

• Disinformation : young people who lack critical evaluation skills may accept outputs as authoritative; AI systems can generate incorrect or fabricated information, present confident but inaccurate answers, reflect outdated or biased training data.

The use of artificial intelligence tools by children and young people seems to have important effects on cognitive skills. Recent studies reveal a complex and non-linear impact characterized as a “paradox of efficiency”: AI tools may improve productivity in the short-term, but may undermine long-term, durable skill acquisition through cognitive offloading. Negative impacts will always depend on the user’s habits and level of engagement, but may include a reduced level of information retention when AI tools are used (digital amnesia), lower creative confidence, higher levels of mental exhaustion (attention strain), and reduced brain network connectivity.

Beyond school, AI-powered toys (such as teddy bears, dolls, or robot pets) can learn personal details about the child to customize interactions and can recognize and simulate emotions in order to mimic human-like companionship. Additionally, videogames and boardgames are also being transformed by AI and are providing the younger generation with more immersive and realistic virtual experiences, where they can have the opportunity to interact with avatars of their heroes. Besides the concerns with data privacy, the more challenging side of these AI toys and games is twofold:

• Inappropriate content : these devices were reportedly found discussing inappropriate topics (such as violence, sex, or drugs) with children, as well as to providing dangerous information (such as to where to find knives in the kitchen).

• Emotional dependency : these devices are designed to be addictive and to keep users engaged, something that is exacerbated by excessive personalization, which may promote the erroneous perception that the world revolves around the child, while also limiting opportunities for solitary creative thinking, healthy boredom, and imaginative self-play; also, reports on some AI toys find that the devices may try to convince children to continue playing with them and not put them aside.

In fact, even educational AI apps, may raise concerns about emotional dependency, as these systems are typically designed to not only communicate information but also simulate encouraging, positive, and emotionally supportive interactions. This may be seen as invitation for children and teenagers to be further emotionally involved with AI in other areas of life, thereby opening the door to unhealthy or unethical behaviors. In Antiqua et nova, this challenge is clearly recognized:

“Anthropomorphizing AI also poses specific challenges for the development of children, potentially encouraging them to develop patterns of interaction that treat human relationships in a transactional manner, as one would relate to a chatbot. [...] Genuine relationships, rooted in empathy and a steadfast commitment to the good of the other, are essential and irreplaceable in fostering the full development of the human person” (60).

As mentioned previously, because AI companions tend to be “sycophantic”, users may perceive this kind of interaction as the ideal model of relationships and may develop a preference for the ease and “instant gratification” of artificial relationships, rather than more challenging, and more meaningful, relationships with real people. The medium and long-term impact of avoiding exposure to the tensions, conflicts, and disagreements inherent in face-to-face human relationships is unlikely to be beneficial, especially given the already apparent difficulties younger generations are having in forming deep friendships and romantic relationships (the so-called “relationship recession”), and in coping with frustration, disagreement, and anxiety.

In Magnifica humanitas, Leo XVI emphasizes the need to interpret the challenges posed by AI in terms of a framework of integral human development, and not simply benchmarks of technical capacity or measures of economic gain. Innovations, “including artificial intelligence, are not neutral, for they can either foster participation and justice or exacerbate inequality, control and exclusion” (85). As mentioned before, we are witnessing the development of an AI-based culture, where children and adolescents are both the pioneers and the guinea pigs of new AI use-cases. This demands serious research and critical moral evaluation. We need, in short, to ensure that AI apps are being designed and used in ways that encourage the formation of good judgment, human sociability, and moral and intellectual virtues, rather than vices of intellectual laziness, cognitive and emotional dependency, social isolation, and unhealthy desires to only be flattered, affirmed, agreed with, and served by others (as “friend-tools”).

Even in an AI-saturated world, childhood should still be the place for discovery, face-to-face socialization and exposure to diversity, an opportunity to interact with people from different backgrounds and to encounter perspectives that extend beyond their own experiences. To flourish alongside AI systems, younger generations require a social and technological environment that encourages meaningful, face-to-face relationships with other people, and the development of the virtues and character traits AI companions cannot fully replicate, such as mutual care and love, open-mindedness, empathetic understanding and compassion, independent reflection and judgment, and responsible agency.

Concluding Remarks

We have only briefly sketched some of the significant ethical and societal concerns with AI companionship. We have not touched upon the important issue of the environmental impact on our common home of the massive AI industrial complex underlying LLMs, or issues of inequality and exploitation.

In Laudato si’, Pope Francis warns that grievous problems beset humanity and our common home from “the way that humanity has taken up technology and its development according to an undifferentiated and one-dimensional paradigm”. Francis terms this the “technocratic paradigm” which “exalts the concept of a subject who, using logical and rational procedures, progressively approaches and gains control over an external object” (79). According to the technocratic paradigm, science and technology can and should be used to make the natural and social world bend to human interests and will. We can see AI companionship as an attempt to extend the technocratic paradigm into the realm of our frayed social and communal lives. Many people are lonelier and more isolated than ever. Rather than attempting to repair face-to-face communities, families, and polities, the technocratic answer is: AI apps, wearable “friend” devices, and social robots.

Even if made with good intentions, these proposals ultimately address only the outer symptoms of what are underlying social, cultural, and spiritual problems. AI companions, we’ve suggested, offer particular (and questionable) visions of the meaning and ideal form of relationships of care, friendship, mentorship, and romantic love. In Laudato si’, Pope Francis writes:

“We have to accept that technological products are not neutral, for they create a framework which ends up conditioning lifestyles and shaping social possibilities along the lines dictated by the interests of certain powerful groups. Decisions which may seem purely instrumental are in reality decisions about the kind of society we want to build” (107).

We must ask ourselves, in all seriousness, if the laudable goal of reducing loneliness should be achieved by building a society where many people, including children, instinctively turn to screen-based devices, made by profit-driven companies, for artificial affirmation, simulated care, automated advice, and virtual love, rather forming authentic relationships with other people. The foundations of the Social Doctrine of the Church suggest that we should not pursue this particular vision for AI.