Webinar series: Research software in the age of Generative AI
Building on the outcomes of the Research Software Engineering in the Age of Generative AI: Building a Community Vision workshop, convened by ReSA and Schmidt Sciences in Edinburgh in March 2026, ReSA is collaborating with organisations and initiatives across the international research software community on a series of public webinars exploring the opportunities and challenges of generative AI (GenAI) for research software.
The workshop identified key opportunities and risks across nine themes and proposed approximately 50 potential activities to advance the safe, reproducible, and effective use of GenAI in research software and research workflows. These outcomes are detailed in the workshop report.
The webinar series aims to broaden participation in these discussions, facilitate the international exchange of experiences and good practices, and identify opportunities for further collaboration.
Upcoming Webinars
AI Guidance for Research Software Engineering Team Leaders
10 November 2026 | 15:00–16:00 CET / 14:00–15:00 UTC
As AI becomes more pervasive in research software engineering, new challenges emerge for team leaders across scientific domains, raising questions such as: How do you ensure high-quality research software while still protecting your developers’ agency? And how do you navigate ethical and technical concerns related to AI practice in research software?
Organised in collaboration with the EVERSE project as part of the EVERSE Network webinar series, this webinar will provide practical guidance to research software managers and open science team leaders on the appropriate and sustainable use of AI in their projects.
Michael Sparks (University of Manchester) will introduce new practical AI guidance developed by EVERSE in the Research Software Quality Toolkit (RSQKit), highlighting the importance of RSE expertise and evaluation skills. Faruk Diblen (Netherlands eScience Center) will share practical tips for RSE managers, drawing on his work on rseng-agent-skills, which packages research software engineering practice for use by AI coding agents. The session will be facilitated by Carlos Martinez-Ortiz (Netherlands eScience Center).
Related workshop theme: Developing a playbook for RSE managers and open-source software project leaders.
When Green Does Not Mean Right: Specifying, Verifying, and Scaling Evaluation of AI-Assisted Research Software
30 November 2026 | 15:00–16:00 UTC
This talk explores why validation of AI-assisted research software has to stay anchored in the science. The reason is simple. AI tools tend to make the signal go away rather than fix the problem, so explicit scientific invariants matter: they give the AI something it can’t silently change. It then turns to specifying behaviour separately from implementation. Executable specifications preserve what matters through regeneration and refactoring - if and only if the checks come from a person or a different system. The AI should not mark its own homework. Finally, it considers how evaluation must scale with AI intensity. Line-by-line review gives way to invariants, reproducible checks and adversarial review, yet designing those checks still rests on human expertise. Manual practice at different levels of AI autonomy remains a key part of the evaluation infrastructure.
Presented by Michael Sparks, chaired by Oscar Seip, and organised by event manager Kyro Hartzenberg, this session is part of the Institute for Research Software’s AI Series, taking place from 24 November to 3 December.
Related workshop theme: Verifying and validating research software.
Understanding Future Incentives Around Publishing, Preserving, and Crediting Software
Date to be confirmed
The goal of this session is to explore publishing, preserving, and crediting open-source and research software. The current academic landscape disincentivises publishing software as opposed to other research outputs, rewards throwaway vibe-coded scripts over well-engineered and robust frameworks, and limits software citations to salient, charismatic projects.
Facilitated by Richard Littauer (CURIOSS), the session will feature three consecutive 15-minute group discussions highlighting aspects of each of these areas, sharing best practices, and exploring ongoing efforts. Each group discussion will begin with a short presentation introducing the topic. Participants will also be able to collaboratively share notes before and after the session.
A recording and transcript of the introductory presentations will be made available following the webinar. The small-group discussions may not be recorded.
Related workshop theme: Understanding future incentives around publishing, preserving and crediting software.
Registration details coming soon.