There is a joint project of DFKI (German Reserach Center for Artificial Intelligence) and VW on human-robot collaboration in manufacturing.
https://www.dfki.de/web/news/dfki-cebit-2017/ihrc
I believe the project to cover especially the technical side of that collaboration. Does this cover the kind of collaboration you have in mind? Or do you think of organizational or other aspects of human-AI interactions?
@ Jan Ole Berndt: Thanks a lot for the links. I was thinking about approaches related to performance measurement of AI-Human colloboration as a comparison of human performance vs. AI performance vs. human-AI-collaboration performance.
Dominic Loske - This is indeed an interesting question which covers several research areas. There has been a lot of work when it comes to potential problems of Human-AI collaboration, beginning with human-computer-interaction in the context of expert systems. Somewhat newer work I am aware of covers the issue of trust into systems and their capability of adapting to a user's needs in order to enhance productivity (especially in the interesting context of "forgetting" systems that attempt to alleviate the problem of "information overload" which can hamper such collaboration). Here are a few examples:
Article A Psychonic Approach to the Design of a Cognitive Companion ...
Article Dangerous Liaisons: Trust, Distrust, and Information Technol...
Thielsch, M. T., Meeßen, S. M. & Hertel, G. (in press). Trust and Distrust in Information Systems at the Workplace. PeerJ.
And, if I may add some shameless advertising of our own work, we are currently developing and using an integrated experiment platform to measure exactly what you are looking for, namely comparisons between human and AI performance in teamwork as well as joint performance of hybrid Human-AI teams:
Conference Paper Challenges of simulating teamwork in organizational scenarios
We have used this platform to generate some interesting findings when comparing human teamwork with that of artificial agents. To summarize it briefly, humans fare better in situations with sparse knowledge being available while artificial agents outperform humans drastically when it comes to even rather simple logical reasoning. You can find our results here:
Conference Paper Simulating Psychological Experiments: An Agent-Based Modeling Approach
Together with colleagues from business, work, and organizational psychology, we are currently working on experiments on hybrid Human-AI teams. However, it will still take some time until we have published results.
Thanks a lot Jan Ole Berndt. Related to the issue of trust into systems I want to add Article Automation and artificial intelligence in business logistics...