Can the Technology That Isolates Us Reconnect Us?
How CGU faculty and students are exploring AI’s paradoxical role in combating the loneliness epidemic.
Is it possible that the same technology being blamed for exposing the cracks in our social fabric might also hold the key to repairing it?
Over the last decade, social media algorithms have been accused of replacing meaningful interaction with endless scrolling. Screen time has displaced face-to-face conversation. Echo chambers have narrowed our exposure to diverse perspectives. AI-powered recommendation systems have been optimized for engagement over connection, keeping us glued to screens while pushing us further apart.
Yet as loneliness reaches epidemic proportions, particularly among young people like graduate students trying to navigate the isolating demands of advanced research, artificial intelligence is also emerging not just as the problem, but as a potential solution.
The paradox is not as contradictory as it may appear. As with most things in life, the issue has never been technology itself, but rather how it is designed, deployed, and managed. The same computational power that has been optimized for engagement and profit can also be harnessed to foster genuine human connection.
At CGU’s Center for Information Systems & Technology, faculty and students alike have been researching the different ways that AI could potentially address loneliness by detecting isolation before it becomes critical, facilitating meaningful connections between people, and providing compassionate support when human connection is unavailable.
Detection and Early Intervention
The most promising applications of AI in combating loneliness may be those that operate quietly in the background, identifying warning signs before a crisis emerges.
Natural language processing algorithms can detect subtle shifts in communication patterns. Things like changes in word choice, sentence structure, or interaction frequency, any of which could signal increasing social withdrawal. While these systems can’t replace human judgment, they could serve as an early detection system that alerts counselors, advisors, or student services professionals to individuals who might benefit from outreach.
Of course, ethical considerations should be taken seriously. Using AI to spot students who might be struggling socially comes with real questions and consequences worth thinking through.
“One’s personal information is at stake,” says Dr. Samir Chatterjee, the Fletcher Jones Chair of Technology Design & Management at CGU’s Center for Information Systems & Technology. “To flag students who may be struggling, AI systems have to track something — browsing habits, how often someone participates in class, even the words they use in messages or posts. Students deserve to know exactly what’s being collected about them, who can see it, and what it will be used for. Without that transparency, helpful monitoring can quietly slide into surveillance.”
AI systems make educated guesses about how someone is feeling based on their behavior, but those guesses aren’t always equally accurate across different populations. What looks like disengagement to an algorithm might just be a difference in how someone communicates. When the AI guesses wrong, real students could pay the price.
“That’s why AI should flag possibilities for a human to look into, not hand down verdicts on its own,” Chatterjee adds.
Gauri Parnaik, an MA student in Applied Cognitive Psychology: User Experience, emphasizes that building trust with these systems comes down to user perception.
“When it comes to loneliness or anything related to mental health, privacy really has to be the top priority,” she says. “A lot of trust comes down to the experience itself — how the system behaves, how transparent it is, and how it communicates its role. From a UX and cognitive psychology perspective, the more invisible and inferential the system becomes, the more cognitively threatening it feels, even if it’s accurate.”
On university campuses, where mental health resources are often stretched thin, these tools could help staff allocate support more effectively. Rather than waiting for students to self-report distress, institutions could proactively reach out to those showing early signs of isolation. The challenge is making sure that such systems enhance rather than erode trust.
Facilitating Human Connection
One of the most direct ways AI could address loneliness is by lowering the barriers that prevent people from connecting in the first place.
Intelligent matching systems can identify shared interests, complementary research areas, or compatible working styles to form study groups, collaborations, or social communities. For international students navigating language barriers and cultural adjustment, AI-powered translation and conversation practice tools could reduce the anxiety that often prevents initial social contact.
But what makes an AI-facilitated connection feel natural rather than algorithmic?
“Meaningful interaction with AI depends less on technological sophistication and more on how systems are designed to support emotional understanding,” explains Chatterjee, whose lab recently built ClareAI, an empathetic virtual companion designed to address loneliness. “Humans respond socially to technologies when those systems demonstrate attentiveness, continuity, and appropriate emotional responsiveness. The goal is not to make AI pretend to be human, but to design interactions that acknowledge users’ feelings and respond in ways that feel supportive and respectful.”
Three design elements were particularly important in the ClareAI project. First, emotional attunement. The system detects cues in language or tone and responds with empathy rather than generic replies. Second, relational continuity. Remembering prior interactions helps build a sense of ongoing conversation rather than isolated exchanges. And third, embodied social presence. A human-like avatar and synchronized expressions can make interactions feel more attentive and engaging.
Companionship and Support
The most pressing danger is creating substitution.
If AI companionship becomes preferable to human connection (more available, more predictable, less emotionally demanding), we risk deepening the very isolation we set out to solve.
But there are contexts where AI companionship can be particularly beneficial. In Chatterjee’s research with ClareAI, findings indicated that AI can be valuable in situations where access to immediate social support is limited, such as graduate students working late hours, individuals living alone, or people adjusting to new environments. In these moments, an AI companion can provide responsive conversation, emotional validation, and a sense of presence that helps reduce feelings of isolation.
“AI companionship can also be helpful as a supplement to human relationships, especially during transitions or stressful periods,” says Chatterjee. “In contexts such as relocation, academic pressure, or temporary social disruption, AI can serve as an accessible conversational partner that encourages reflection and emotional expression.”
The shift toward digital companionship has become particularly visible among younger generations. Since the COVID-19 pandemic, researchers have observed that Gen Z individuals are socializing differently than previous generations. Younger adults are going out less frequently, forming fewer in-person friendships, and spending more time in digital environments. Naturally, that means some are increasingly turning to AI-based conversational companions as a form of interaction when they feel isolated or stressed.
The most thoughtful research treats AI companions less as replacements for human relationships and more as tools for building the capacity to form them.
“I don’t see AI itself as inherently crossing that line,” says Parnaik. “The issue is what role it starts to occupy in a person’s ecosystem of support. If AI is used to extend thinking, reduce friction in tasks, or support reflection, it stays in the support category. But when it starts to replace reciprocal human relationships — especially the kind that involve mutual vulnerability, accountability, and emotional risk — that’s where it becomes problematic.”
She also stresses the importance of distinguishing AI from therapy. “Therapy is not just ‘talking to someone who responds well.’ It’s a structured, ethically bound, clinically trained relationship with clear responsibility, boundaries, and escalation frameworks. I think it’s critical that users understand AI is not a substitute for therapy, both from a safety and expectation standpoint.”
“When designed with empathy, relational continuity, and clear psychological boundaries, AI companions can function as a supportive bridge helping individuals feel heard and emotionally supported while still encouraging meaningful connections with other people,” Chatterjee explains.
These technologies should be evaluated not just by their immediate effects but by their long-term consequences. Does the tool help someone eventually form human relationships, or does it reduce the motivation to do so? Does it provide temporary support during a difficult period, or does it create dependency?
The Choices Ahead
Loneliness is not a problem any single discipline can solve. The questions raised by AI companionship cut across computer science, psychology, ethics, public health, and education, and they ultimately come down to something more fundamental: how we want to live together, what we value in human connection, and what we are willing to trade away in the name of convenience.
That breadth is part of what makes this moment difficult. The technology already exists. Ignoring it is not a neutral choice. But neither is deploying it without asking what we want it to do, who it might fail, and what safeguards should be in place before it scales.
The ClareAI project, led by Dr. Samir Chatterjee with Dr. Nayana Bose of Scripps College and graduate students Armin Abazari and Parzon Faridani, is one attempt to work through those questions in practice, supported by a BLAIS Challenge grant for intercollegiate research across The Claremont Colleges.
The technology that isolates us can also reconnect us. But only if the people building it understand that the goal is not to replace human connection. It is to clear the path toward it by removing barriers, recognizing those in need, and offering support where none would otherwise exist.
Whether that vision holds depends on the choices being made right now, while these systems are still taking shape and still open to direction.