Know Center Research

Human-AI Interaction

What impact will the increasing use of AI and machine learning have on humans? Will humans and AI compete with each other? Or will we succeed in making the best possible use of the strengths of both sides and enhancing them through conscious cooperation between humans and machines?

Human-AI Interaction

Research Focus

The Human-AI Interaction research area works on interactive machine learning techniques, immersive and visual analytics, and AI-driven computer user interface methods that promote mutual understanding and cooperation between humans and AI. For this to succeed, both humans and AI must be able to communicate with each other, provide feedback, and act on that knowledge. The result for us is the use of an AI that enables fluid cooperation between humans and machines and gives us the freedom to use our human strength of genuine creativity.
Human-AI Mixed Initiative Interaction

This research area develops novel human-in-the-loop methods (interfaces, models, and systems) that enable humans to obtain information about AI decisions and provide feedback. Likewise, AI will be enabled to solicit human input to improve the performance of the human-machine team.

Explainable AI

The goal of this research direction is to develop explainable algorithms and to provide the user with methods for interpreting the AI’s learning processes and reasoning. Explainable AI plays a key role in trustworthiness and acceptance of AI. Research focuses on algorithms for computing explainable features, methods for validating and analyzing them, and methods for explaining these features in an understandable form.

Human-Aware AI Models

Our goal is to develop intelligent systems that can “reflect” their actions in terms of human perception, intentions, and other, contextual factors. We are interested in collecting data and generating data-driven models to understand what people’s affective and emotional states are in different interactions with technology and how they influence the interaction. Such systems should be able to adapt to different human states while generating models of human intentions and behaviors.

Potential

New possibilities for interaction

AI applications can only develop their full potential if the interaction with people is also designed to meet human needs. Conversely, we are also convinced that the trustworthy AI of the future will also help unleash new human potential. Namely, when the right information is delivered at the right place, at the right time, and thus social processes such as decision-making, multi-stakeholder dialogs, or co-creative development processes in dislocated, multilingual teams are supported in the best possible way.
Media Crawling & Analysis

We can scan across platforms and using text recognition in relevant digital libraries, (social) media and company portals. Domain-specific information is extracted by the AI. Appropriate visualization tools support the analysis.

Monitoring Platform

A visual user interface to search, display and explore themes, trends, geo-specific patterns, correlations, and interrelationships. It can be adapted very precisely to the specific requirements of the user, depending on the information needs.

Modelling tools

Advanced optimization of AI-assisted text analysis through semantic model management, user feedback input, and text annotations (additional text information such as annotations).

Data Insights

Have patterns in time series data recognized by AI (events, anomalies) and inspected and examined with visual interfaces.

Search in time series databases

Finding similar past results in very large databases to compare and learn from.

Transparent AI

We give subject-matter experts the opportunity to pass on knowledge to the AI with little effort. It is then possible to understand and validate AI decisions. Reasons for incorrect results can be identified and eliminated.

Intelligent user guidance

A “user-aware” AI guides the user, depending on his/her analytical goals, through various data analysis steps

custom-tailored colLAborATiOn visualizaTion

Collaborative text annotation, discussion, and evaluation of results by multiple users. Design and implementation of specialized, use case and requirement-specific visualization components. Extensibility of the framework (Visualizer) supports rapid prototyping of visual analytics solutions.