Grounded Copilot: How Programmers Interact with Code-Generating Models
Grounded Copilot: How Programmers Interact with Code-Generating Models
复制标题
DOI：
10.1145/3586030
复制
发表时间：
2022-06
影响因子：
--
通讯作者：
Shraddha Barke;M. James;N. Polikarpova
中科院分区：
文献类型：
--
作者：
Shraddha Barke;M. James;N. Polikarpova
在代码生成的模型的最新进展中，像Github Copilot这样的AI助手承诺将永远改变编程的面孔，但我们提出了第一个基于程序员与Copilot互动的扎根的面孔。 20名参与者 - 使用助手的一系列经验 - 他们解决了我们的主要发现的潜水员编程任务。助手是双峰的：在加速模式下，程序员知道下一步要做什么，并在探索模式下使用副驾驶来更快地到达那里。用于提高未来AI编程助理的可用性。
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.