Project description
The interdisciplinary project investigates the co-creation process of humans and AI in the context of creating programme code. The focus is on the design of trustworthy interfaces for the use of co-generators in programming education and professional software development.
Two target groups are considered: (1) programming beginners, especially computer science students in their first semesters, and (2) professional software developers. When using code generators in the context of a computer science degree programme, there is a risk that relevant programming skills will not be acquired. When used in professional software development, the central question is whether code generators can efficiently support the development process or whether their use is associated with increased debugging effort.
The interdisciplinary project combines AI methods and approaches from experimental cognitive research. Quality measures for generated code are investigated and further developed and approaches are designed and implemented that enable a combination of code generators with inductive programming to augment generated code.
Psychological studies and experiments are used to analyse the effects of different human-AI interfaces on skill acquisition and appropriately calibrated confidence in generated code. The use of code generators such as Copilot in professional software development will be analysed via surveys regarding application contexts and subjective evaluations along performance criteria. The knowledge gained in the project is intended to contribute to the evidence-based design of trustworthy co-creation processes in the field of programming.
The aim of the project is the empirical investigation and evidence-based design of trustworthy co-creation processes for the generation of programme code.
Project team

Prof. Dr. Ute Schmid
Member of bidt's Board of Directors and the Executive Commitee | Member of the Bavarian AI Council | Head of Cognitive Systems Group, University of Bamberg
