Tricky${}^2$: Towards a Benchmark for Evaluating Human and LLM Error Interactions

Cole Granger, Dipin Khati, Daniel Rodriguez-Cardenas +1

Large language models (LLMs) are increasingly integrated into software development workflows, yet they often introduce subtle logic or data-misuse errors that differ from human bug…