SocialGenPod: Privacy-Friendly Generative AI Social Web Applications with Decentralised Personal Data Stores

TL;DR

Unlike centralised Web and data architectures that keep user data tied to application and service providers, it is shown how one can use Solid - a decentralised Web specification - to decouple user data from generative AI applications.

Abstract

We present SocialGenPod, a decentralised and privacy-friendly way of deploying generative AI Web applications. Unlike centralised Web and data architectures that keep user data tied to application and service providers, we show how one can use Solid -- a decentralised Web specification -- to decouple user data from generative AI applications. We demonstrate SocialGenPod using a prototype that allows users to converse with different Large Language Models, optionally leveraging Retrieval Augmented Generation to generate answers grounded in private documents stored in any Solid Pod that the user is allowed to access, directly or indirectly. SocialGenPod makes use of Solid access control mechanisms to give users full control of determining who has access to data stored in their Pods. SocialGenPod keeps all user data (chat history, app configuration, personal documents, etc) securely in the user's personal Pod; separate from specific model or application providers. Besides better privacy controls, this approach also enables portability across different services and applications. Finally, we discuss challenges, posed by the large compute requirements of state-of-the-art models, that future research in this area should address. Our prototype is open-source and available at: https://github.com/Vidminas/socialgenpod/.

Figures (5)

• Figure 1: Centralised architecture (left) makes it difficult for users to have control over their data and use their data across applications. Decentralised architecture (right) decouples data from applications, giving users control over their data and making it easier for them to use their data across apps.
• Figure 2: Example Web application data flow in SocialGenPod. Alice allows Bob to interact with or query her personal data through a Web app. The app uses a retrieval service to find permitted documents in Alice's Solid Pod that are most relevant to Bob's query. The app then uses one of the available generative AI models to generate a contextualised response for Bob.
• Figure 3: RAG with private data in SocialGenPod. A chatbot without RAG (top) is unable to answer a user's query. The same chatbot with RAG (bottom) produces a better response.
• Figure 4: Login and configuration menu in SocialGenPod.
• Figure 5: Full chat interface in SocialGenPod, showing chat threads and available models.