The Software Engineering research community produces hundreds, if not thousands of papers every year, and yet the impact of most studies remains limited. In this talk I discuss why this might be so, namely, a lack of direction. The SE community is quite well prepared in its methodological designs (though improvements can be made here too), but not so well in the creation of a cumulative body of knowledge. The SE community tends to focus very strongly on the “topic of the day”, and as a community we chase every topic we can find, without grounding these topics in prior literature. I posit that this constant stream of publications that remain disconnected is due to a lack of attention and appreciation for theory building. In this talk I will illustrate the problem using examples, and offer a potential way forward to address this. The solution offered will not be straightforward to implement, but if successful, we can strengthen the SE literature and make it more relevant for both research and practice.

Software engineering has always been a story of tools and teamwork. Today, both are being redefined. As organizations experiment with hybrid and remote models, and as generative AI reshapes how we write, review, and reason about code, we face a dual transformation: of work and of intelligence. This talk explores the five paradoxes of modern software work—balancing productivity, engagement, innovation, belonging, and well-being—through evidence from multiple empirical studies across companies worldwide. It connects these insights to the emerging use of GenAI in development, asking: what happens when not only where we work but how we think changes? The keynote invites researchers and practitioners to look beyond efficiency metrics and engage with the deeper design of intelligent, intentional workplaces that actually work.

Opening + 5 papers

Advanced Artificial Intelligence (AI) techniques — such as deep learning, foundation models, agentic systems and multiagent systems — are increasingly used for improving automation across the software lifecycle. While these techniques may accelerate software production, their inherent characteristics, including autonomy, opacity, generative capabilities, and non-determinism, may impose understudied or unmapped risks to both software quality and engineering teams. Understanding and managing these risks is particularly critical in contexts that demand strict guarantees around dependability, performance and developer well-being. Bringing together experts in complementary research topics, this panel aims to address four questions:
1. Understudied Factors: Which software quality and human factors impacted by AI remain understudied? How are they possibly impacted?
2. Unmapped Risks: What are the less-understood risks that characteristics of advanced AI techniques impose on software quality and human factors?
3. Underexplored Opportunities: Which capabilities of these techniques (and why) should be better leveraged to further improve software quality and team work?
4. Longitudinal Research: How can we design and conduct rigorous long-term studies to better measure AI impact on software maintenance, evolution and developer well-being?
Longitudinal Research: How can we design and conduct rigorous long-term studies to better measure AI impact on software maintenance, evolution and developer well-being?

6 papers

3 trabalhos

Software quality can be understood from multiple perspectives: the quality of the products we build, the processes through which we build them, and the experiences of the people who use and develop software. This talk explores these perspectives through the research journey of the USES Research Group, from early work on usability and user experience to research on requirements engineering, software architecture, and human factors in software engineering.
Beyond research results, this journey also highlights another dimension of quality: fostering researchers. Over the years, USES has become not only a place for investigating how to build better software and better support software development, but also an environment where new generations of Software Engineering researchers can grow and thrive. Today, former USES researchers hold faculty positions at universities across four of Brazil’s five regions, where they continue to conduct research and mentor new researchers. Drawing on selected research contributions and lessons learned along this journey, the talk reflects on how a human-centered view of Software Engineering can extend from the software we build and the people who build it to the researchers who will shape the future of the field.

(Educação em Engenharia de Software)

5 papers (ENGLISH)

Interactive Discussion & Reaction Session: 2-3 industry practitioners/sponsors share 5-minute reactions on AI governance & guardrails, followed by open audience Q&A. (ENGLISH)

2 papers (ENGLISH)

4 papers