<b>Enhancing Human-AI Interactions: The Impact of Pseudo-code Engineering on Improving Predictability and Stability in Large Language Models</b> - Appendix A

This research focuses on an innovative approach to enhancing the predictability and stability of interactions between humans and Large Language Models (LLMs), such as ChatGPT, through the use of pseudo-code. By investigating three distinct interaction formats (natural language, a hybrid of natural language and pseudo-code, and exclusive pseudo-codo) this study utilizes a mixed-methodology approach that combines qualitative content analysis with quantitative analysis methods like standard deviation and coefficient of variation.Key findings from this study indicate that exclusive use of pseudo-code in interactions with ChatGPT increases the predictability and stability of the model's outputs by 11%, leading to responses that are more detailed and logically structured. On the other hand, prompts that integrate both natural language and pseudo-code see a 20% increase in content richness compared to those that use only natural language. This suggests that a structured, logical format, developed through pseudo-code engineering, significantly improves the effectiveness of human-LMM interactions.The implications of this research are substantial, suggesting that the principles of pseudo-code engineering could be beneficial not only for ChatGPT but also for other generative LLMs like Claude, Gemini, and Llama. The application of these findings could potentially transform GenAI-assisted activities across various sectors including education, customer support, and more, thereby advancing strategies in human-computer interaction.