Name of the participant: Michael Chromik
Description of the IT-research project: Intelligent systems based on methods of artificial intelligence (AI) and machine learning (ML) have achieved considerable automation potentials in many areas. Despite these advances, their use in sensitive areas of practice is limited, because information relevant to decision-making is often only available as implicit knowledge in the minds of decision-makers. In these areas, an iterative interaction between humans and AI systems is the more promising approach (e.g. in the semantic analysis of text documents).
There are two challenges for such “hybrid intelligence” systems:
(i) User-centered design: Compared to traditional digital systems, AI systems are non-deterministic and change over time. Besides the optimization of algorithmic accuracy, the focus is therefore on the alignment with user needs;
(ii) Explanability: Novel AI systems work as black box systems, i.e. often not even the developers know how the systems arrive at their results and predictions.
The aim of the project is to explore and prototypically implement interaction concepts for collaborative processing and authoring of textual content between domain experts and AI-based assistance systems in the legal domain. Current methods of Natural Language Processing (NLP) and Natural Language Generation (NLG) are combined with the human-in-the-loop approach of Interactive Machine Learning (IML). A holistic interaction concept will be developed, which structures the entire process and guides the user through it with the help of an artificial assistant. Emphasis is put on the design of user interfaces for situations with a multitude of transitions and iterations between user and AI system.
Software Campus partners: LMU München, DATEV
Implementation period: 01.03.2020 – 31.05.2021
