Research Study Title: Evaluating and Comparing Artificial Intelligence Explainability Systems
You are being invited to take part in a research study. In order to help you decide it is important for you to understand why the research is being done and what it will involve. Please take time to read the following information carefully. You may wish to discuss it with others. For any further information or questions about my research, please contact me on: jonathan.frawley <at> durham.ac.uk or contact my supervisor on christopher.g.willcocks <at> durham.ac.uk
Artificial Intelligence (AI) systems are increasingly becoming a part of everyday life. How these systems operate is hidden from most users, however. This study aims to evaluate a variety of systems which explain the inner workings of Artificial Intelligence systems to end users. We will do this by looking at assessing how interpretable different explainability systems are across different models and different problems.
We ask for some personal information at the beginning of the study, this is so that we can evaluate how explainability systems affect different segments of the population. None of this info can be used to identify you personally. Because of this, we are also not able to delete an individual user's data once it has been submitted. You are free to decide whether or not to take part. You have the right to withdraw at any point prior to submitting your personal information on the following page. Please note that once you submit this information by clicking the submit button, that your information can not be deleted from that point on. We store information even for partially filled-out surveys, and these cannot be deleted and will form part of our research.
Data will be stored on personal servers and computers of lead researcher and collaborators. Data will also be stored on Durham University's Microsoft OneDrive system for backup and collaboration purposes. Data may also be made publicly available and shared for the purposes of future studies. This data will be used mainly for the purposes lead researcher's PhD research but will also be used for academic publications.
We are aiming to get a large sample size for this study. A bigger sample size will improve the robustness of the outcomes of the study. You have not been specifically targeted for this study. By participating, you will be providing us with valuable information that will give us insight into how to assess the quality of AI explainability systems.
We will present you with a variety of different tasks, for example recognising everyday items in images and reading comprehension. For each task we will present you with all or a subset of the following information:
- Input (always shown)
- AI model output
- AI model reasoning explanation
We will then ask a series of questions about the information presented. Please give your best guess as to what you think the output should be, even if you are not sure. Note that some of the problems are intended to be very difficult.
We will provide you with descriptive text for each problem.
This survey will take about 10-15 minutes to complete.
If you wish to retain a copy of this consent form study, please either save this page or contact me on: jonathan.frawley <at> durham.ac.uk and a copy of this consent form will be made available to you.
By clicking the "Start" button below, you are confirming that you that the following statements are correct:
- I confirm that I have read and understand the information about the above-named project and have had the opportunity to ask questions.
- I am aged 18 or over.
- I understand that participation is voluntary.
- I agree to take part in this project.