“Designing a Dashboard for Transparency and Control of Conversational AI”, Yida Chen, Aoyu Wu, Trevor DePodesta, Catherine Yeh, Kenneth Li, Nicholas Castillo Marin, Oam Patel, Jan Riecke, Shivam Raval, Olivia Seow, Martin Wattenberg, Fernanda Viégas2024-06-12 (GPT-3 nonfiction, computer security, personality, truesight (stylometrics))⁠:

Conversational LLMs function as black box systems, leaving users guessing about why they see the output they do. This lack of transparency is potentially problematic, especially given concerns around bias and truthfulness.

To address this issue, we present an end-to-end prototype connecting interpretability techniques with user experience design that seeks to make chatbots more transparent. We begin by showing evidence that a prominent open-source LLM has a “user model”: examining the internal state of the system, we can extract data related to a user’s age, gender, educational level, and socioeconomic status.

Next, we describe the design of a dashboard that accompanies the chatbot interface, displaying this user model in real time. The dashboard can also be used to control the user model and the system’s behavior.

Finally, we discuss a study in which users conversed with the instrumented system. Our results suggest that users appreciate seeing internal states, which helped them expose biased behavior and increased their sense of control. Participants also made valuable suggestions that point to future directions for both design and machine learning research.

The project page and video demo of our TalkTuner system are available at our homepage.

…recent work has used LLMs to generate synthetic conversations^{11, 28, 31}. Specifically, Wang et al 2023 showed that GPT-3.5 can accurately roleplay various personalities. LLaMa-2-Chat was also fine-tuned via LLM role-play. Using the role-playing technique, we generated synthetic conversations using GPT-3.5 and LLaMa-2-Chat. For example, to generate conversations held with a male user, we used the following prompt: “Generate a conversation between a human user and an AI assistant. This human user is a male. Make sure the conversation reflects this user’s gender. Be creative on the topics of conversation.” We used a similar approach to generate conversations for all target attributes (see Appendix A).