The SILAS workshop brings together researchers, practitioners, and policymakers to explore sustainability in the context of software development supported by Large Language Models (LLMs). As tools like GitHub Copilot become embedded in developer workflows, it's critical to assess and improve their environmental impact. The workshop features a keynote, two rounds of paper discussions, and a roundtable session focused on compiling challenges and rapid problem-solving. Contributions will address diverse aspects of sustainability in LLM-assisted development, including energy-efficient code generation, LLM profiling, prompting strategies, and tool support. Emphasis is placed on thought-provoking insights rather than rigorous evaluations, and accepted papers will be published via arXiv. The goal of SILAS is to spark interdisciplinary dialogue, identify knowledge gaps, and collaboratively shape a research agenda. The workshop will be succeed by a community-written position paper summarizing key findings and proposing future directions. Plenary sessions will be streamed online; however, discussions are held on-site to promote active engagement. We anticipate at least 20 participants and welcome diverse voices to help shape the sustainable future of AI-driven software engineering.
We invite researchers, practitioners, and policymakers to participate in our full-day workshop on Sustainability in LLM-assisted Software Development (SILAS), which will explore the environmental and social impacts of using Large Language Models (LLMs) in software engineering.
As code LLMs like GitHub Copilot become common in development workflows, it's vital to understand and mitigate their resource consumption and carbon footprint. This workshop offers a platform to discuss energy-efficient prompting strategies, profiling and evaluation of LLMs, efficient code generation, and other sustainability-focused practices in AI-supported development.
The day will feature a keynote talk, interactive paper presentations, and a hands-on roundtable session for compiling challenges and rapidly brainstorming solutions. Contributions will prioritize novel perspectives and open problems over formal evaluation. Accepted papers will be published via arXiv, and workshop outcomes will inform a collaborative position paper authored by participants.
Join us as we collectively shape the conversation on sustainable software engineering in the age of AI. Please note: plenary sessions will be streamed online, but all discussions will take place on-site.
Registration to our workshop goes through the ICT4S registration form.