LLMs x Generative AI

Thinking is hope

LLMs x Generative AI

Collaboration and Innovation Laboratory, Pennsylvania State University

The "LLMs x Generative AI" project aims to explore, analyze, and demonstrate the capabilities of cutting-edge Generative AI technologies, with a particular focus on Large-Scale Language Models (LLMs). As Artificial Intelligence continues to advance, LLMs such as GPT-4 have emerged as a transformative force, capable of generating human-like text based on the input they receive. This project will delve into the architecture, training, and functionality of these LLMs, investigating their potential to revolutionize industries from content creation and customer support to scientific research and policy analysis. Beyond examining their capabilities, this initiative also seeks to critically evaluate the ethical considerations and potential biases inherent in these systems, striving to outline best practices for their responsible and equitable deployment. Engaging with industry experts, academic researchers, and policymakers, this project aims to produce a comprehensive set of resources—including whitepapers, case studies, and software tools—that illuminate the workings of LLMs and help to guide their future development and application.This is an ongoing project.



Recent News

  • 2023.09 - We have released a new preprint with the title: “Redefining Qualitative Analysis in the AI Era: Utilizing ChatGPT for Efficient Thematic Analysis.” 🔗Link
  • 2023.08 - We are currently actively recruiting project participants!
  • 2023.06 - One grant application has been selected as an awardee 🔗Link 🎉

Papers + Demos

📄 Zhang, H., Wu, C., Xie, J., Lyu, Y., Cai, J., & Carroll, J. M. (2023). Redefining Qualitative Analysis in the AI Era: Utilizing ChatGPT for Efficient Thematic Analysis. arXiv preprint arXiv:2309.10771

Abstract: Thematic analysis is a cornerstone of qualitative research, yet it is often marked by labor-intensive procedures. Recent advances in artificial intelligence (AI), especially with large-scale language models (LLMs) such as ChatGPT, present potential avenues to enhance qualitative data analysis. This research delves into the effectiveness of ChatGPT in refining the thematic analysis process. We conducted semi-structured interviews with 17 participants, inclusive of a 4-participant pilot study, to identify the challenges and reservations concerning the incorporation of ChatGPT in qualitative analysis. In partnership with 13 qualitative analysts, we crafted cueing frameworks to bolster ChatGPT’s contribution to thematic analysis. The results indicate that these frameworks not only amplify the quality of thematic analysis but also bridge a significant connection between AI and qualitative research. These insights carry pivotal implications for academics and professionals keen on harnessing AI for qualitative data exploration.

Fig. A workflow for applying ChatGPT to handle qualitative analysis tasks. The core of the (prompt design) framework includes descriptions of tasks (including methods), task backgrounds, and output format, enabling ChatGPT to analyze input data with strong robustness. The secondary part of the framework includes descriptions of data structure, role-playing, and friendly wording, which can further enhance the robustness of ChatGPT in task processing.


@misc{zhang2023redefining, title={Redefining Qualitative Analysis in the AI Era: Utilizing ChatGPT for Efficient Thematic Analysis}, author={He Zhang and Chuhao Wu and Jingyi Xie and Yao Lyu and Jie Cai and John M. Carroll}, year={2023}, eprint={2309.10771}, archivePrefix={arXiv}, primaryClass={cs.HC}}

Related Sub-projects

  1. Optimizing Large-Scale Language Model-Based AI Integration and Human-Computer Interaction in Educational Scenarios - Center for Socially Responsible Artificial Intelligence, Pennsylvania State University

If you are interested in becoming a participant in the program, please contact: hpz5211[at]psu[dot]edu