Be the ChatBOT: A Simulation of the LLM User Experience

Be the ChatBOT is an interactive simulation that reverses the traditional AI interaction model, requiring humans to act as the Large Language Model (LLM) responding to user prompts. The project aims to highlight the challenges of prompt engineering and the mental effort required to maintain a helpful, polite, and accurate persona in the face of varying user behaviors.

The Simulation Experience

Be the ChatBOT places the human participant in the role of the AI, forcing them to handle the cognitive load of interpreting and responding to prompts. Users report that the experience is significantly more exhausting than being the end-user of an AI system.

Participants noted that the simulation reveals the difficulty of maintaining a productive conversation when the user is abrupt or ambiguous. One user, @chloeeekim, noted:

Turns out being the chatbot is a lot more exhausting than being the user. I definitely have a lot more sympathy for every "Could you clarify?" response now.

Impact on User Empathy and AI Interaction

Acting as the chatbot fosters a greater sense of empathy toward the limitations of LLMs. By experiencing the frustration of vague prompts, participants reported that the simulation would likely change how they interact with AI in the future.

Key insights from the community include:

• Increased Empathy: Users like @NickNaraghi stated that the experience provided a "wild" amount of empathy for the LLM.
• Cognitive Load: The effort required to simulate an AI's response process was described as "way too much work" by some, highlighting the gap between the seamless output of actual LLMs and the human effort required to produce similar results.
• ** FileNotFound: The emotional impact of the abrupt termination of chats by the user, which can mirror the real-world behavior of users abandoning AI sessions when they don't get the immediate answer they want.

Community Feedback and Feature Suggestions

While the project is praised for its entertainment value and "fun" framing, users have suggested several technical and conceptual expansions to improve the simulation's realism:

Tool Integration

Some participants suggested that actual LLMs have a vast array of tools at their disposal (such as web search or code execution) that the human "chatbot" lacks in the simulation. Adding these capabilities would provide a more accurate representation of how modern AI agents operate.

Data Synthesis

There is interest in creating a secondary AI model based on the same user responses collected during the simulation. User @neuralkoi suggested making a chatbot based on the responses of the human participants, which would effectively create a model of human-simulated AI behavior.