Grounded Copilot: How Programmers Interact with Code-Generating Models
Powered by recent advances in code-generating models, AI assistants like Github Copilot promise to change the face of programming forever. But what is this new face of programming? We present the first grounded theory analysis of how programmers interact with Copilot, based on observing 20 participants—with a range of prior experience using the assistant—as they solve diverse programming tasks across four languages. Our main finding is that interactions with programming assistants are bimodal: in acceleration mode, the programmer knows what to do next and uses Copilot to get there faster; in exploration mode, the programmer is unsure how to proceed and uses Copilot to explore their options. Based on our theory, we provide recommendations for improving the usability of future AI programming assistants.
Wed 25 OctDisplayed time zone: Lisbon change
11:00 - 12:30 | |||
11:00 18mTalk | Grounded Copilot: How Programmers Interact with Code-Generating Models OOPSLA Shraddha Barke University of California at San Diego, Michael B. James University of California at San Diego, Nadia Polikarpova University of California at San Diego DOI | ||
11:18 18mTalk | Turaco: Complexity-Guided Data Sampling for Training Neural Surrogates of Programs OOPSLA Alex Renda Massachusetts Institute of Technology, Yi Ding Purdue University, Michael Carbin Massachusetts Institute of Technology DOI Pre-print | ||
11:36 18mTalk | Concrete Type Inference for Code Optimization using Machine Learning with SMT Solving OOPSLA Fangke Ye Georgia Institute of Technology, Jisheng Zhao Georgia Institute of Technology, Jun Shirako Georgia Institute of Technology, Vivek Sarkar Georgia Institute of Technology DOI | ||
11:54 18mTalk | An Explanation Method for Models of Code OOPSLA DOI | ||
12:12 18mTalk | Optimization-Aware Compiler-Level Event Profiling OOPSLA Matteo Basso Università della Svizzera italiana (USI), Switzerland, Aleksandar Prokopec Oracle Labs, Andrea Rosà USI Lugano, Walter Binder USI Lugano Link to publication DOI | ||