Work

Designing Cognitive ‘Copilots’: Augmenting Our Minds Through Prompting Conversational Interactions

Year: 2024
Type: dissertation
Abstract: Natural language user interfaces (NLUIs) have become increasingly popular for a range of tasks and activities, given the interactive, adaptive, and contextual interactions they can offer. While the goal of many NLUIs is to make it easier for the user to request, find, or generate information or content to complete certain tasks more efficiently, it is argued here that they can also be effective at scaffolding people’s thinking while performing cognitive tasks. It is hypothesised that the way this can be achieved is by proactively asking people questions to engage them in reflective thinking about a task at hand. The focus in this thesis is on openended tasks that benefit from such reflective thinking, as it can enable people to discover alternative perspectives, approaches, and possibilities which can help to progress with the task. The NLUIs presented in this thesis are embedded into the interfaces used to perform different types of cognitive tasks and were thus named ‘cognitive co-pilots’. The tasks included a collaborative exploratory data analysis task, a complex decision-making task, and a creative three-dimensional drawing task. This PhD research explored the opportunities, as well as the challenges of ‘embedding’ such prompting co-pilots into these types of cognitive activities. It also examines the design parameters at the interface, such as the adequate timing, phrasing, and delivery of prompts – and how they can affect ongoing activities. The thesis reports five studies conducted on different types of proactive NLUIs and provides a novel way of conceptualising them in terms of how they can extend human cognition.
Language: English
Open Access status: green
OpenAlex ID: W6981081348
Cites: 435
Cited by:
Related to: 9