‘BhaavaBot’ detects isolation and offers empathetic conversation, companionship and mental wellness support
Remember Theodore from the 2013 film Her—the lonely writer who finds comfort in an intuitive AI companion? While the movie was futuristic, loneliness like his is now a common reality in today’s fast-paced, hyper-connected world.
In response, two students from Dhirubhai Ambani University (DAU) have created an AI chatbot that not only detects a user’s loneliness but also listens, chats and suggests ways to overcome isolation.
Originally a classroom project, the initiative has grown into a multidisciplinary effort. The prototype, called BhaavaBot – The Feeling Machine, has been developed by MSc Data Science students Vraj Patel and Meet Panchal under the guidance of Prof. Anupam Rana.
Derived from the word bhaav (emotion), the chatbot offers non-judgmental companionship, entertainment through music and movie suggestions, and instant emotional support via real-time conversations. Prof. Rana said the bot is trained on thousands of empathetic dialogue samples and even includes an animated talking avatar.
The app, a mental wellness support tool, will be available to users once the AI undergoes further refinement. A pilot version is currently ready for testing.
Trained on 1.15 lakh conversations
“The BhaavaBot dataset includes 1,15,000 conversation pairs designed to train an AI model for loneliness support,” Prof. Rana explained. “It contains 1,528 emotion categories, with loneliness accounting for 24% of them.”
He said loneliness typically affects three groups:
- teenagers who interact less with family and peers but stay active online;
- students and professionals who relocate and live away from home;
- elderly people experiencing empty-nest loneliness.
“The prototype works, but due to the sensitivity of the topic, we are expanding the dataset to make the chatbot more effective and more trustworthy,” he said.
Students have also created a charming animated avatar that nods, moves and makes conversations feel more human, he added.