Establishing AI-based research (artificial intelligence)
We did not train AI; we established a new way of working in research. (artificial intelligence)
Telekom researchers work with large volumes of interview material every day, but producing secondary analyses required considerable manual effort. The goal was not simply to demonstrate the use of AI, but to establish an operating model that accounted for corporate data security requirements, research quality expectations and the available technology environment. To achieve this, we adapted the capabilities, limitations and potential applications of existing AI tools to Telekom’s own research processes. (artificial intelligence)
The outcome of the project was not training material, but a repeatable way of working. The research team learned how to produce faster, more reliable AI-supported secondary analyses without compromising research quality or data security. (artificial intelligence)
As part of the programme, we tested the LLMs used by Telekom alongside other research-support platforms, compared their outputs with previous human analyses, and developed the recommended prompting and validation methodology. The training was followed by a workshop, enabling the team to apply the new way of working confidently in their own projects. (large language model)