➡️ Do you have the feeling that you have got into an argument with someone you like? Take out your phone and record the conflict with "It's not about winning". Take your time for a first round of self-reflection. Once the conflict is over, you can use the feedback from our app to reflect on your argument!
Inspiration
Our inspiration for "It's not about winning" comes from the realization that many disputes, whether personal or professional, escalate due to miscommunication and unaddressed emotions. We understand that good communication is not just about the words exchanged but also involves self-reflection and a willingness to see things from different perspectives. The very existence of our tool serves as a prompt for this kind of introspection. By additionally offering a neutral, AI-driven platform for conflict analyzation, we hope to encourage individuals to reconsider their approach to conflicts, promoting a more reflective and empathetic way of interacting. Our aim is to shift the focus from winning an argument to understanding and resolving the underlying issues, fostering healthier and more productive conversations.
What it does
We have designed our platform to transform disputes into more productive and goal-oriented conversations. This goal is achieved by analyzing audio files from conflicts, detecting the words and phrases used alongside the volume of the speech. Based on this analysis, the application provides tailored communication advice aimed at an understanding of communication mistakes and fostering a more constructive dialogue. But most importantly it encourages self-reflection before, and afterwards. This will help to improve future conflicts and conversations.
How we built it
Our application, "It's not about winning," was crafted leveraging the robust capabilities of OpenAI's GPT-4 for its core functionality. We chose AI, specifically GPT-4, for its impartiality – it acts like a neutral observer, offering unemotional and balanced communication advice. This AI-driven approach ensures unbiased assistance in managing and resolving conflicts.
To create an intuitive user interface, we have utilized a no-code tool. This approach allowed us to focus more on the application's functionality (i.e. we have done intensive research on conflicts, how they work, what can be done wrong, etc.) and less on intricate coding for the user interface, streamlining the development process.
The backend of our application is powered by Google Cloud APIs. We are utilizing Google Cloud Functions for our backend processes, ensuring efficient handling of our application's operations. These functions work in tandem with Google's Speech-to-Text API, which we employed for its experimental real-time support and speaker diarization, allowing our application to analyze conversations as they happen.
Another important feature of our application is the real-time volume meter. This tool provides users with visual feedback on the volume of their conversation. We recognized that volume is a critical aspect of communication, especially in conflict situations, and this feature helps users to be more aware of their tone and volume, encouraging an even more constructive dialogue.
Our application runs as a web application and should be able to run on all devices with a web browser!
Challenges we ran into
A first hurdle was achieving consistent output in terms of timestamps and JSON format from GPT-4. This was crucial for a good user experience. So we had fun prompting GPT to get a good and consistent output format 🧐
Another major technical challenge we faced was real-time speaker diarization, which remains a quite complex task in the field of audio processing. Accurately distinguishing between different speakers in a live conversation and analyzing each speaker's tone, volume, and content in real time presented us with a number of difficulties that we could not fully resolve in 24 hours (with our limited expertise in this field).
Accomplishments that we're proud of
The extensive research we conducted on conflict situations and effective communication strategies! This deep dive into the corresponding literature not only enriched our understanding of conflict dynamics but also significantly enhanced our ability to craft more effective prompts for GPT-4. And this significantly improved "It's not about winning".
What we learned
Throughout this project, we learned a great deal about the complexities of human communication and the nuances of emotional expression. We have also learned how to build a useful application that uses Generative AI very, very fast.
What's next for It's not about winning
The next step for "It's not about winning" is to enhance its capability to handle realtime conflict-situations. Also there is still a lot of things to do when it comes to improving GPT-4 prompts, so that the model acts even more like a neutral advisor (one that also works for more complex conflict situations, understanding everyone's individual way of disputing, sarcasm, etc.).
Built With
- google-cloud
- gpt4
- nocode
- speech-to-text
Log in or sign up for Devpost to join the conversation.