Inspired by the recently founded "Think Tank KI" of the University of Bonn, colleagues at the HRZ—whose usual responsibilities are already related to AI—have connected in an internal focus group under the name HRZ AI-Lab for the first time.
It is the objective of the HRZ AI-Lab to combine the existing knowledge of the HRZ staff about AI and respective expertise in core areas of AI, in order to support and actively enrich the work and development of the Think Tank AI.
The meetings and events within the context of the HRZ AI-Lab are meant to give every interested employee at the HRZ the opportunity to discuss the development and use of AI-machines and to gain first practical experiences.
First hackathon on 20 June 2024
On 20 June 2024, the first internal "hackathon" on the topic of AI took place as part of the HRZ AI Lab.
Initial experiences with large language models (LLMs) were gained during the event. For this purpose, virtual machines were set up in advance and the challenges and hopes for a possible productive operation at the HRZ were recorded. During the hackathon, various tasks were tested, including the operation of own models, the possible management of models, the strengths and weaknesses of user interfaces and the use of own databases. In the second step, the connection of GPUs is planned, which not only accelerate computing power, but also make some AI applications such as computer vision possible in the first place.
Scientific basis of AI
In preparation for the hackathon in the context of the HRZ AI-Lab the scientific background of an AI was explored. Three key areas were identified:
- Machine Learning (ML): An aspect of AI that is concerned with the developments of algorithms and techniques to allow computers to learn from data and make predictions.
- Neuronal Networks: Models which are inspired by the structure and functionality of the human brain and can be used for the processing of complex data patterns.
- Data science: The use of statistical methods and algorithms to analyse and interpret large amounts of data and their processing and use for the methods of AI.