After extensive experimentation with ChatGPT in my operations research work, I've identified six genuinely useful applications and three where it falls short. The results may surprise you.
6 Recommended Uses
Code Engineering
1. Code Templates. ChatGPT generates reasonable starter implementations for common problem types, providing a foundation you can build upon rather than starting from scratch.
2. Debugging. The tool can identify errors in code samples, though you should verify fixes independently as bugs may persist.
3. Understanding Existing Code. ChatGPT explains previously written code effectively, helping you comprehend unfamiliar implementations or legacy systems.
Dealing with People
4. Explaining Technical Concepts. ChatGPT creates initial drafts translating complex technical ideas into accessible language suitable for non-technical audiences, though refinement remains necessary.
5. Data Request Emails. The tool generates professional follow-up messages efficiently, producing ready-to-send templates with minimal editing required.
6. Conference Talk Descriptions. ChatGPT creates compelling promotional text for conference presentations, offering solid first drafts needing only minor adjustments.
3 Uses to Avoid
Mathematical Model Building
1. Concept Refreshers. While ChatGPT provides textbook-like definitions, it lacks deeper understanding, functioning like a student memorizing content rather than comprehending it.
2. Algorithm Examples. ChatGPT produces significantly flawed illustrations—I've found superior feasible solutions through visual inspection alone.
3. Comparing Modeling Approaches. Responses lack necessary detail and rigor for meaningful comparative analysis of optimization techniques.
The Bottom Line
ChatGPT excels at mechanical writing and coding tasks but remains inadequate for the sophisticated mathematical reasoning central to OR research. Use it where it helps, but don't expect it to replace domain expertise.
Written by
Jonasz Staszek