Please Gamble Responsibly: Training to Protect Research Integrity in the Age of AI-assisted Coding

Slide deck and transcript of a talk delivered at the German Research Software Engineering Conference 2026 in Stuttgart, Germany. Title: Please Gamble Responsibly: Training to Protect Research Integrity in the Age of AI-assisted Coding Abstract: Using generative AI tools to generate source code or to augment a programmer's own code is becoming an ever more widespread approach to software development. While evidence of associated gains in software engineers' productivity is far from settled, little doubt remains that large language models are capable of generating relatively extensive volumes of source code that will compile and run. If we accept the assumption that this technology will remain widely available in the future, its existence has deep implications for research, including for the development and maintenance of Research Software. This talk will explore some of the open questions surrounding the use of LLMs for coding in research, including: What role should Research Software Engineers take in an environment where it has never been easier for researchers to produce code? What threats does genAI pose to the integrity of the research process?What benefits might genAI have for researchers and RSEs who can harness the technology?How can we counter the hyped-up claims of LLM developers and simultaneously avoid falling into denialism? What must researchers and RSEs learn about the use of LLMs to produce (Research) Software? In this context, the talk will briefly review existing literature and summarise ongoing efforts of a working group of Carpentries community members to produce new "AI Carpentry" workshops. The Carpentries is a global community of practitioners teaching essential software and data skills for research. The primary goal of the new workshops is to help researchers and RSEs make informed decisions about whether and how to use LLMs in the development of software. With separate workshops targetting novice programmers and researchers/RSEs who are already able to code, the group aims to help learners make safer, more effective use of LLM coding assistants. The workshops will also build learners' awareness of the short- and long-term risks posed by these approaches to the independence and ability of the individual as well as to the integrity of the research. Offering a nuanced perspective on the subject matter, this talk will be most relevant to RSEs and other digital Research Technical Professionals who want or need to teach others -- formally or informally -- about the development and use of Research Software. It will provide details about how to get involved with the development and delivery of these new workshops, and offer participants an opportunity to give feedback on the project.