| Departments |
Computer Science, Zienkiewicz Institute for Modelling, Data and AI
|
|---|
| Description | This project will explore how the process of combining human and algorithmic inputs in decision-making is affected by the number of agents involved. It will require an online platform and recruitment of sufficient participants to provide robust evidence. The study should attempt to emulate a high-stakes decision context that a non-specialist can make and product safety recall decisions are suggested, but not required. It is anticipated that the decision agent will be a Wizard of Oz system. The central concern is to evidence the difference, if any, between three different conditions. A human user making decisions alone; doing so with the help of a single agent; and with assistance from several agents. The research aims to illuminate the possible significance, advantages and drawbacks of multi-agent support in combined decision-making.
Methodology
• Literature Review: Examine current research combined decision-making, multi-agent systems (other than robots) and CSCW.
• Platform Development: Create an online decision-making study platform able to present problems, incorporating display of, and possibly interaction with, multi-agent contributions and able to record decision responses along with general feedback.
• User Studies: Design and conduct experiments to evaluate decision-making quality and consistency under different conditions.
• Data Analysis: Employ mixed-methods analysis to assess the effectiveness of different multi-agent combination and interaction approaches.
• Guideline Development: Synthesize findings to create design guidelines for algorithmically-supported decision processes. |
|---|
| Preparation | Something you will learn on this for this project.
• Decision Modelling: Develop and refine a set of challenging, consequential decision scenarios that require users to choose between competing objectives.
• Programming: Develop or adapt an online user-study system to capture sufficient responses across multiple treatments to allow robust statistical analysis.
• Data Science: Collect and analyze user study data to derive insights on human-algorithm combined decision-making.
• Human-Computer Interaction: Design intuitive interfaces and multiple interaction modalities, with a focus on interactions with intelligent algorithms.
• Computer-Supported Cooperative Work: Investigate different combination models between humans and different numbers of assistive agents.
• Evaluation Methods: Develop metrics and methods for measuring decision quality and trade-offs in human-algorithm combination. |
|---|
| Project Categories |
Artificial Intelligence (AI), Human Computer Interaction (HCI) |
|---|
| Project Keywords |
Machine Learning |
|---|
Level of Studies
|
| Level 6 (Undergraduate Year 3) |
yes
|
|---|
| Level 7 (Masters) |
yes
|
|---|
| Level 8 (PhD) |
yes
|
|---|
