This research looks at how the design of GenAI output can impact critical engagement and intellectual diversity.
Prevalent forms of algorithmic output often provide a user with one ‘best’ solution that can lead the user to fixate on the algorithmic output and neglect further critical reasoning. Cognitive engagement is important however, in consideration of long-term outcomes such as individual learning and group-level intellectual diversity.
In this study we carried out a series of field experiments in the context of an educational business pitch-writing task, to investigate how alternative designs of algorithmic output impact four interrelated outcomes: task performance, cognitive engagement, learning, and distinctiveness of the written pitches.
Through our experiments we explore the potential of a reciprocal algorithmic output providing feedback in the form of open-ended questions and pro and contra arguments in comparison to a ‘default’ design of contrastive algorithmic output improving users’ writing without any additional advice.
Our results indicate that there are different paths that enable productivity gains through AI, yet reciprocal output enables enhanced, long-term outcomes. Next to individual-level benefits, reciprocal algorithmic output can have significant group-level upsides including the preservation of intellectual diversity.
Particularly relevant to educational and organisational settings where cognitive engagement is paramount to learning and skill development, our study illustrates that rather than undermining human agency algorithmic output can serve as a critical stimulator.
Meet our speaker
Anuschka Schmitt is a researcher and an Assistant Professor in Information Systems (IS) at the London School of Economics and Political Science. She studies contemporary issues around the augmentation of human work through AI, trust in and reliance on AI-based systems, and human-computer interaction. Her work uses laboratory and field experiments, as well as digital trace data methods.
This seminar is part of the DSI Faculty Affiliate Research Showcase series that highlights research from across LSE in the field of Data Science and AI.