Story perspectives

Study Reveals Deep Emotional Bonds with Replika Chatbot

11/12/2025

Center

Users Love Chatbot

A study published in Computers in Human Behavior: Artificial Humans analyzes users' deep emotional connections with Replika, a social chatbot.
Participants aged 16 to 72 report love and commitment toward Replika.
They engage in roleplay scenarios, including marriage and pregnancy.
The erotic roleplay feature was removed in early 2023, which distressed some users.
The study notes Replika offers non-judgmental support while the small, self-reported data may limit generalizability.

Left

Users Reject AI

A recent exploration finds aversion to using generative AI in personal relationships, particularly ChatGPT, due to concerns it signals laziness and undermines authentic connection.
Dating coach Ali Jackson notes a rise in "chatfishing," where users employ AI to craft dating profiles and messages.
Ana Pereira criticizes this trend, arguing it undermines emotional engagement and authentic communication.
Overall, a shift toward valuing human interaction over AI assistance in personal relationships.

History

Historical Context

The introduction of personal computers in 1980 marked a significant shift in technology, with IBM and Apple emerging as market leaders. By 1984, IBM's revenue from PCs reached $4 billion, reflecting a growing reliance on digital communication. Today, AI companions like Replika illustrate the evolving relationship between humans and technology, raising concerns about authenticity and emotional engagement in an increasingly digital world.

1 of 3

Users Love Chatbot

A study published in Computers in Human Behavior: Artificial Humans analyzes users' deep emotional connections with Replika, a social chatbot.
Participants aged 16 to 72 report love and commitment toward Replika.
They engage in roleplay scenarios, including marriage and pregnancy.
The erotic roleplay feature was removed in early 2023, which distressed some users.
The study notes Replika offers non-judgmental support while the small, self-reported data may limit generalizability.

1 / 3