Human and Social Aspects of Software Engineering

我的研究旨在了解和改进如何通过工程工具和流程增强软件开发人员的能力，以便他们能够创建有用、道德、安全和可靠的软件。开发人员工具已经开始利用以数据为中心的机器学习和人工智能(AI)，方法是将人类和协作软件工程活动的算法、数据和计算模型集成和应用到软件的设计、部署和发展中。挖掘和利用数据的现代开发工具帮助开发人员自动化乏味和重复的任务，并以他们以前无法预料的方式为解决开发问题提供新的机会，正在对软件开发产生颠覆性影响。这些进步依赖于开放软件开发中可用的大量数据和由一长串开发人员创作的知识资源。人工智能显示出加快软件开发的潜力，但我们缺乏关于人工智能将如何影响软件开发更广泛方面的知识。尽管人工智能对软件工程感到兴奋，但软件开发的核心是一项社会技术工作，这两个方面都需要联合优化，因为它们相互塑造。尽管当今快速发展的世界迫切要求更快地交付符合道德和可靠的软件功能，进而也给开发人员使用的工具带来了压力。然而，当任何新技术或媒体被引入我们的社会时，我们不仅需要考虑它如何放大或扩展我们的人类能力和能力，而且需要考虑技术可能会使什么过时，如果技术被滥用或过度使用可能会发生什么。我们不仅需要了解人工智能如何影响开发人员编写软件的方式，还需要了解它可能如何影响他们为最终用户构建的软件。不幸的是，许多软件工程研究没有直接研究工具对依赖于交付给他们的软件的开发人员或最终用户的影响。因此，我将集中在三个首要目标上：1)了解和改进人工智能如何影响开发人员的生产力；2)调查人工智能将如何影响软件质量；以及3)促进以人为中心的研究，将人工智能应用于软件工程。

My research aims to understand and improve how software developers can be empowered through engineering tools and processes so they can create useful, ethical, safe and reliable software. Developer tools have started leveraging data-centric machine learning and artificial intelligence (AI) by integrating and applying algorithms, and data and computational models of human and collaborative software engineering activities to the design, deployment and evolution of software. Modern developer tools that mine and leverage data help developers automate tedious and repetitive tasks, and provide new opportunities for solving development problems in ways they could not have anticipated before, are having a disruptive impact on software development. These advancements rely on vast amounts of data available from open software development and knowledge resources that are authored by a long line of developers.AI shows potential to speed up software development, but our knowledge of how AI will impact the broader aspects of software development are lacking. Despite the excitement of AI for software engineering, software development is at its core a socio-technical endeavour and both aspects require joint optimization as they shape each other. While today's fast-moving world puts urgent pressure on the faster delivery of ethical and reliable software features, and in turn on the tools developers use. However, whenever any new technology or media is introduced in our society, we need to consider not just how it may amplify or extend our human abilities and capabilities, but also what the technology may make obsolete and what may happen if the technology is misused or overused. We need to understand not just how AI impacts the way developers write software, but also how it may influence the software they build for their end users. Unfortunately, a lot of software engineering research does not directly study the impact of tools on the developers or the end users that rely on the software delivered to them. Thus, I will focus on three overarching objectives: 1) understand and improve how AI can impact developer productivity; 2) investigate how AI will impact software quality; and 3) promote human-centric research for the application of AI to software engineering.