The problem Dwight solves

Current AI-driven chat platforms lack effective monitoring, leading to unchecked spread of harmful content and regulatory breaches. Manual intervention is slow and inefficient, while compliance with regulations remains challenging. Users engaging in inappropriate behavior often go unaddressed due to limited intervention mechanisms and collaboration features. This perpetuates a culture of impunity and undermines user trust.

The Dwight project addresses these issues by offering real-time monitoring, proactive risk detection, and automated compliance enforcement. By intervening promptly, providing guidance, and facilitating collaboration, Dwight ensures responsible use of generative AI-based chatbots, fostering safer and more ethical online environments while preserving user trust and upholding regulatory standards.

Challenges we ran into

Throughout the development of our Dwight, we encountered various challenges in both implementing the chatbot and creating the accompanying website. Integrating llama Guard for real-time monitoring proved complex, requiring extensive customization and optimization to align with our project's architecture. Technical hurdles arose with the OpenAI Python SDK, particularly in fine-tuning algorithms for real-time risk detection and intervention, impacting project timelines. Additionally, integrating together.ai presented challenges due to its decentralized cloud services and complex APIs, necessitating collaboration with the development team and causing implementation delays. Limited resources and support compounded our difficulties, hindering progress and impacting timelines. Moreover, adapting to evolving APIs and technologies posed ongoing challenges, requiring frequent updates and modifications. These obstacles were present not only in creating the chatbot but also in developing the website, resulting in a challenging but ultimately rewarding development process.