What’s Does the Recent Research Say about AI-Ships?

An informal review of recent research reflects AI companions anti-social nature and potential impact on communities.

Excitement around this year’s two-week Wimbledon tournament ran high, with ESPN reporting it as the most-watched edition in six years and averaging 721,000 viewers across 250 hours of coverage. As a young person, I tuned in annually for big matches. My friends and I would gather on community courts to play (awfully), while also reenacting some of the McEnroe antics during those summer months. This summer, social interaction is in transition, as fewer young people run to the local tennis courts to whack a ball around. Instead, a growing number are turning to AI companions for social and romantic engagement through sites like Candy.ai. In one month, Candy.ai drew more than 10 million visits (~80% from young men aged 18–24) underscoring this shift in socialization away from real-life interactions.

AI companionship, or AI-ship, has rapidly moved from niche experimentation into the mainstream. Platforms such as Candy.ai, Joyland, GPTGirlfriend, and Replika attract millions of monthly visits, largely from young men between the ages of 18 and 24. Framed as digital partners, these services promise intimacy on demand: conversation, validation, even sexualized interaction. For many, the appeal lies in a world where loneliness is rising, and real-life relationships often feel fraught or inaccessible.

The central tension, however, is stark. While AI-ship can provide temporary relief from feelings of isolation and a low barrier to entry into companionship, it also risks reinforcing dependence on artificial intimacy. Recent literature suggests that rather than guiding users toward healthier in-real-life (IRL) relationships, these tools may distort expectations of reciprocity, erode trust, and weaken ties to families, peers, and communities. These studies point to both fleeting comfort and longer-term risks, particularly for young men navigating identity, mental health challenges, and social connection in this rapidly expanding corner of human-technology interaction.

AI companions may offer a sense of connection in moments of need, but their widespread adoption carries troubling implications for men’s wellbeing and for the vitality of communities built on authentic human relationships.

Loneliness & Connections

Loneliness among young men has emerged as a defining public health concern. The Real Talk report highlights that rates of mental ill-health among young men in Australia rose from 26% in 2007 to 32% in 2020–22, with loneliness and social disconnection cited as key drivers (Bulluss & Baker, 2025). This study also shows that young men often struggle with help-seeking, influenced by norms of self-reliance and stoicism that discourage reaching out. The Movember Young Men’s Media Landscaping Report underscores how boys turn to online spaces for belonging, with 88% playing video games weekly and high proportions engaging with podcasts and social media influencers (Fisher et al., 2025). This gap often emerges when offline networks fail to provide the support or intimacy that young men need. In these moments, digital influencers and AI companions appeal by offering guidance, constant availability, and a sense of validation that feels easier and more immediate than human relationships.

AI companions provide immediate comfort and validation, often filling an emotional void for young men. In interviews with Replika users, Pan and Mou (2024) identified two dominant and somewhat conflicting ways users described their interactions. First, in the discourse of “idealization,” users framed the AI companion as a perfect partner capable of offering unconditional support, never arguing or disappointing them. In contrast, the users' discourse of “realism” reflected more complex views where users acknowledged the artificial nature of the relationship and its limits, even as they still derived comfort from it. These contrasting narratives reveal how AI-ship can both heighten fantasy and spark reflection about the nature of intimacy, which is central to understanding why these tools appeal to young men while potentially shaping their expectations of real-world relationships.

The same qualities that make AI companions attractive can also entrench dependency. Döring et al. (2025) warn that while AI-enabled partners can gratify users’ needs, they also create risks of emotional dependence. Reynaud (2024) highlights how users may reinforce feelings of rejection or isolation when AI girlfriends become substitutes for real interactions. Taylor (2023) cautions that repeated abusive or demeaning interactions with AI companions can normalize disrespectful habits, which then “leak over into their relationships with humans.” This suggests that AI companionship risks not only isolating users but also shaping unhealthy relationship habits that damage trust and respect.

Adolescents and marginalized young men appear particularly vulnerable to the harms of AI companionship. Downloads of Replika surged during the COVID-19 pandemic, showing how digital companions became substitutes during times of heightened social isolation (Metz, 2020). Additionally, Reynaud (2024) explored how incel communities may use AI girlfriends to reinforce misogynistic ideologies, rather than alleviate rejection. The eSafety Commissioner’s Being a Young Man Online report (2024) found that boys describe digital life as simultaneously liberating and anxiety-inducing, with experiences of pressure, body image concerns, and exposure to harmful gender norms. For many boys, AI-ship may provide momentary companionship, but following established patterns of online engagement, over time it deepens patterns of loneliness, dependency, and disconnection from human peers, especially among vulnerable groups.

AI image of AI companions.

Physical Health Links

Loneliness has been described as a public health risk on par with major physical hazards. The U.S. Surgeon General (Murthy, 2023) equated chronic loneliness to smoking fifteen cigarettes a day, a framing that highlights the potential health risks of AI companionship. The Real Talk report documents that poor mental health in young men is closely tied to loneliness and social disconnection, contributing to negative outcomes such as substance use disorders and suicide (Bulluss & Baker, 2025). These findings highlight that loneliness, and its intensified state through AI-ships, is not merely an emotional challenge but potentially a pathway to negative health and wellness outcomes.

Patterns of overuse in digital technologies add additional stressors for young men. Nawaz et al. (2025) found that excessive smartphone use among Australian men aged 18–40 was strongly linked to increased stress, anxiety, and disrupted sleep, with many reporting an inability to disconnect from constant connectivity. Similarly, Prochnow et al. (2023) demonstrated that excessive internet use and online gaming were associated with heightened risks of depression and anxiety, while healthier engagement in online networks could provide social support. These results suggest that when digital reliance tips into compulsion, it directly undermines both mental and physical wellbeing.

AI-ships introduce new opportunities for digital dependency. Reilama (2024) describes how users of AI companion users often report becoming “hooked” or addicted, indicating behaviors resembling compulsive use. This emotional dependence can exacerbate stress and reduce resilience. Allen (2024) further illustrates this dynamic in a study of Replika users, where the introduction of erotic role play features triggered distress, grief, and feelings of abandonment among those who had come to rely on the AI companion as a coping mechanism. Döring et al. (2025) add that AI-enabled partners, while capable of providing gratification, also risk creating cycles of emotional dependence that destabilize users’ health.

AI image of isolated young man.

The health burden of loneliness may be amplified by AI-ship, which replaces protective social connections with compulsive, tech-mediated substitutes. These substitutes may soothe users temporarily but often leave them more vulnerable to stress, disrupted sleep, and declining mental and physical wellbeing.

Impact on Community & Civic Life

AI companionship often embeds problematic scripts into digital relationships. Reilama (2024) describes how users of AI companion services are given “god-like power” to shape submissive roles for companions, normalizing unequal dynamics and increasing the likelihood of addiction to the service. Similarly, categories available on GPTGirlfriend include coercion-oriented and domination themes, which set expectations of compliance rather than reciprocity. This content often intersects with the broader influencer culture online, where Renström and Bäck (2024) show that manfluencers reinforce misogynistic attitudes, finding that young men who follow more of these influencers are more likely to dehumanize women IRL. These dynamics risk distorting boys and men expectations that controlling or coercive behavior is acceptable, and more importantly anticipated within relationships.

Algorithms on major platforms intensify exposure to harmful gender norms. The Safer Scrolling report demonstrates that TikTok algorithms quadrupled exposure to misogynistic content within five days of use, which then became normalized in both online and offline behavior among young people (Regehr et al., 2024). The eSafety Commissioner’s Being a Young Man Online report (2024) similarly documents how harmful gender norms present in digital environments underpin increased risks of gender-based violence. These findings highlight that AI companionship exists within a wider ecosystem where algorithms accelerate and normalize harmful content.

Parents and caregivers also express deep concern about how digital content shapes boys’ development. ReachOut’s Parenting in the Digital Age survey reports that 59% of parents worry about their children’s ability to cope with online pressures, particularly misinformation and harmful social comparison (2023). Many describe AI or digital relationships as poor substitutes for real-world coping and learning. This intergenerational anxiety underscores how AI-ship not only impacts young men directly but also ripples outward into family and community wellbeing.

Over time, AI-ships could undermine not just personal relationships but also community trust, civic participation, and gender equity.

Conclusion

AI companionship (AI-ship) has rapidly become part of many young men’s daily lives, offering fleeting relief from loneliness but at significant cost. The evidence across reports and studies shows that while digital partners may provide comfort in moments of isolation, they also reinforce unhealthy relationship habits and entrench dependence. These dynamics can distort expectations of intimacy and reduce the likelihood of young men seeking healthier support systems.

The risks are compounded when viewed through the lens of health and community. Research links loneliness and digital overuse to disrupted sleep, stress, anxiety, and depression. Beyond individual health, AI-ship has the potential to undermine civic engagement by embedding misogynistic attitudes, further eroding trust, and weakening ties to families and communities. Taken together, the harms outweigh the temporary solace these systems provide.

Moving forward, the conversation must expand beyond individual use toward societal responsibility. Reports stress the need for policy guardrails, digital literacy education, parental guidance, and community-level interventions to protect boys and young men while fostering authentic relationships. What is ultimately at stake is not just the wellbeing of individuals, but the shape of our communities and the values we choose to cultivate in the age of AI.

What’s Next?

Quite frankly, I feel like an curmudgeon writing this blog on a new technology, particularly 24 hours after having two of my closest friends stay with me for a long weekend. The research is clear on the value of our IRL social support networks (Waldinger, 2023), and the research is growing on the negative impacts associated with AI-ships. Until tech companies design AI companions with some transparency, clear goals toward building real life social support systems, and raise barriers to accessing companions that are driven on their dependency, it’s hard to jump onboard this tech wave. I realize that eventually it may just sweep me away.

This leaves us all (parents, educators, policymakers, or young men) at a moment to engage in shaping healthier pathways forward. AI-ship is not going away, but how we respond will determine the degree of its impact on communities. Get involved: push for stronger digital literacy programs in schools, advocate for platform accountability, and start conversations in your own networks about what authentic connection looks like. By speaking up and acting now, we can push for technology that serves our wellbeing rather than quietly undermining it.

References

• Allen, M. (2024). My AI Companion: An Examination of the Removal of Erotic Role Play. [Unpublished manuscript].

• Bulluss, A., & Baker, D. (2025). Real Talk: Masculinity and young men’s mental health online. Orygen.

• Döring, N., et al. (2025). The impact of AI on human sexuality: A five-year literature review. Computers in Human Behavior, 150, 1-10.

• eSafety Commissioner. (2024). Being a young man online. eSafety Commissioner Report.

• Fisher, K., et al. (2025). Movember young men’s media landscaping report. Movember Foundation.

• Metz, C. (2020, May 6). Who needs friends? Meet a robot. The New York Times. https://www.nytimes.com

• Nawaz, S., et al. (2025). Exploring smartphone usage patterns and perceived dependency: Australia. Journal of Behavioral Addictions, 14(2), 1-15.

• Pan, H., & Mou, Y. (2024). Human–AI romantic relationships: Discursive constructions among Replika users. Computers in Human Behavior, 146, 1090–1091.

• Prochnow, T., et al. (2023). Online gaming network communication dynamics, depressive symptoms, and social support: A longitudinal network analysis. Journal of Behavioral Addictions, 12(3), 272–273.

• ReachOut. (2023). Parenting in the digital age. ReachOut Australia.

• Regehr, C., et al. (2024). Safer scrolling: Addressing harmful content algorithms. University of Toronto Press.

• Reilama, A. (2024). Me, my AI boyfriend, and I. [Master’s thesis, University of Helsinki].

• Renström, E., & Bäck, H. (2024). Manfluencers and young men’s misogynistic attitudes: Perceived threats to men’s status and social dominance orientation. Frontiers in Psychology, 15, 1787.

• Reynaud, E. (2024). Examining interactions between incels and AI girlfriends. [Master’s thesis, Concordia University].

• Taylor, J. (2023). AI will always love you: Studying implicit biases in romantic AI companions. Journal of Digital Relationships, 5(2), 1-12.

• U.S. Surgeon General, Murthy, V. (2023). Our epidemic of loneliness and isolation: The U.S. Surgeon General’s advisory on the healing effects of social connection and community. U.S. Department of Health and Human Services.