Navigating the Future of AI-Augmented Software Engineering: Part 4
AI Summary
Summary: Applying ChatGPT in Vanderbilt Computer Science Department
Context:
- The discussion focuses on the application of ChatGPT to address a challenge at Vanderbilt’s Computer Science Department.
- The problem-solving example demonstrates the potential of prompt engineering and natural language programming.
Problem:
- The department held a faculty retreat to brainstorm improvements in teaching, rankings, research, and student success.
- Faculty members were asked to submit topics they wanted to discuss at the retreat.
Solution:
- Instead of manually categorizing the topics, ChatGPT was used to:
- Organize the topics into six or seven categories.
- Generate headings for each category.
- Summarize the rationale for each grouping.
Process:
- Faculty ranked their interest in each topic area.
- ChatGPT was tasked with mapping faculty to discussion groups based on their interests.
Challenges & Iterations:
- Initial attempts using greedy algorithms and bipartite graph matching led to suboptimal allocations.
- ChatGPT’s first solutions either overfilled some groups or left some faculty unassigned.
- After several iterations and using different algorithms, including randomization, ChatGPT successfully assigned faculty to their preferred topics while balancing group sizes.
Outcome:
- The faculty were impressed with the AI-assisted process.
- The exercise showcased the efficiency of using ChatGPT for problem-solving without manual intervention or coding.
Conclusion:
- The example highlighted the synergy between humans and AI in solving complex problems.
- It reinforced the idea of using AI tools like ChatGPT for problem-solving in educational settings.