Preprint
Aug 2026
Beyond the Traceback: Using LLMs for Adaptive Explanations of Programming Errors
While LLM-rewritten messages significantly improved subjective evaluations, with pragmatic messages rated as clearer and less cognitively demanding, these perceived gains did not translate into statistically significant improvements in objective debugging performance.
Alexandru-Radu Moraru, Shreyan Biswas, U. Gadiraju
· 0 citations