Can We Trust the Model? Reproducibility and Reliability of LLM Results
LLMs are currently used in many areas of database technology. Their areas of application are very diverse and include design tasks, programming, query generation, optimization, data integration, and much more. The results are often remarkably good and surprisingly accurate. However, all application areas share one important limitation: we cannot yet rely on the results of LLMs to be completely trustworthy, reproducible, or stable in the long term.
This means that LLMs are very well suited for building prototypes, but not yet for developing robust, industry-ready systems. This challenge will be the focus of the spring meeting of the GI Database Systems specialist group.
We would like to thank Snowflake for sponsoring the event.
Programm
We start our programme on Wednesday (04.03.) at 13:00. The programme ends on Thursday (05.03.) around 13:00.
Venue
This year's spring meeting will take place on the campus of the Faculty of Informatics and Data Science at Bajuwarenstraße 4, 93059 Regensburg.
Venue