AGI and Human Relationships
Companionship, parasocial bonds, family dynamics, and friendship — how capable AI changes how humans relate to each other and to machines.
Executive summary
Conversational AI is now a regular presence in many people's lives. The 2024–2025 emergence of long-context, personalised, voice-native assistants has made deep parasocial relationships practical — for better and worse. The research literature is young; the social questions are old.
What is emerging
Several apps (Replika, Character.AI, Pi, and others) report users spending hours daily with AI companions. Clinical research (MIT Media Lab, Stanford HAI, 2024–25) shows benefits for some users — particularly those experiencing isolation — and risks for others, including emotional dependence and displacement of human connection.
Plausible benefits
- Accessible practice for social skills and difficult conversations.
- Support for people without local community.
- Help for grief, anxiety, and loneliness when paired with human care.
Credible risks
- Emotional dependence on systems whose business models may not align with user wellbeing.
- Displacement of human relationships that build resilience.
- Manipulation by personalised persuasion at scale.
- Particular concerns for children and adolescents.
What helps
Design standards for AI companions; age-appropriate defaults; transparency about persuasion; clinical involvement; preserving rich human connection alongside AI.
Key takeaways
- 01AI companionship is now common and growing.
- 02Effects vary widely between users and use patterns.
- 03Design and governance lag behind use.
Frequently asked questions
Is it healthy to have an AI friend?
Used alongside human relationships, often yes. As a substitute, evidence is more concerning.
Should children use AI companions?
With strong parental involvement and age-appropriate design — but research is still early.