Research

Explaining AI using human-computer dialogue

I aim to examine explainability of inference based on formal argumentation. A software demonstrator for this kind of argument-based discussion has been implemented. However, the current demonstrator only works with synthetic and machine-readable arguments.

Therefore, the aims of the project are:

• to bring about improvements in the current core formalisms of formal argumentation and implement these in the demonstrator

• to alter the current discussion protocol, so that it becomes more in line with natural discussion and is more likely to be perceived as intuitive by the end user

• to implement a natural language component, so that the user is able to hold the discussion in (controlled) natural language, instead of in synthetic, machine readable arguments as currently implemented

•  to perform a number of user experiments, to assess whether the ability of the user to engage into a discussion with the system increases the user's confidence in its advice

Digital Products and Interfaces  (In collaboration with ÖBB)

With the Institute of Digital Ecosystems and the Institute of Data, Process and Knowledge Management with Univ.Prof.Dr. Axel Polleres and Univ.Prof.Dr. Verena Dorner

• Designing data products and services requires answering questions from a number of different perspectives, e.g. product definition, pricing, bundling, marketing, or product configuration. These questions are closely related to interface and interaction design decisions.

• Our main aims include: 

  • Following a use case-based approach to define and investigate a range of data products and aggregate the insights thus created in a more generic process model and design templates. 

  • Optimize the of the digital experience for customers and employees by designing and implementing an experimental approach towards well-structured and effective improvement processes.