An examination and fine-tuning of ChatGPT for interactive problem solving in support request systems

This thesis examines how Support Request Systems, frequently used by IT departments, can be made more interactive and efficient through the use of large language models like ChatGPT. Within the scope of this study, the ChatGPT model was fine-tuned and tested to simulate a Support Request System, and its performance was analyzed. The fine-tuned model demonstrated success in understanding support requests, classifying them, and providing accurate solutions. The findings reveal that large language models possess significant potential in understanding language and generating responses aligned with user requests. Moreover, it was observed that the model's natural language processing capabilities can accelerate operational processes and enhance user experience. The study also shows that ChatGPT can be effectively utilized not only in IT departments but also in customer support systems and other business processes. In this context, the flexibility of language models and their applicability to various domains offer a promising approach for future implementations. In conclusion, this thesis evaluates the impact of large language models on support systems, proving that ChatGPT's interactive problem-solving capabilities provide significant advantages compared to traditional methods.