The Ethics Assessment of research proposals now takes a great deal of time. To change this, Dennis Reidsma, chair (and previously member) of the Computer & Information Science Ethics Committee, designed the 'pre-ethics tool'. Soon, you will be able to receive tailor-made advice before submitting your proposal simply by completing a questionnaire. "This means that the plans that we receive will be more complete, the quality will be higher, and everyone will save a lot of time."
Reidsma, who is an Associate Professor of Human Media Interaction, designed the prototype for the tool together with three students. The proof of concept will be complete by the end of 2022, at which point other faculties and departments can readjust the tool for use within their own disciplines. Reidsma and the students devoted a great deal of time towards making the tool user-friendly, and it can therefore be easily tailored to suit highly specific disciplines. Due to the immense value that the tool will provide to society, the Executive Board awarded Reidsma the Juliana Medal for designing the tool.
Facilitating designs for vulnerable target groups
Reidsma enthusiastically explained that the pre-ethics tool also makes it much easier for researchers and students to conduct research into vulnerable target groups. "For example, it can be used for people with literacy problems, children, or people with a mental or physical disability. The questionnaire means that in exchange for very little effort, researchers and students can gain rapid insight into the rules, tips and tricks concerning research into specific target segments, including vulnerable groups. This removes an important barrier and I hope that in the future, it will enable more products to be designed for people in vulnerable groups."
Research strategy amplifies or minimises ethical risks
Reidsma explains that before conducting research, you must always examine whether it is ethically responsible and which rules apply. "Human Media Interaction very often involves human subjects, and in such cases, your research proposal must undergo a mandatory evaluation by the Ethics Committee (EC). With other research projects, evaluation by the EC may be desirable even if it is not mandatory. For example, the development of Artificial Intelligence (AI) also involves ethical risks. If you do not train the algorithm well enough it can function in a discriminatory manner. The research strategy that you adopt can play an important role in amplifying or minimising these types of risks."
Ethical risks involved in Human Media Interaction research
Human Media Interaction research involves a variety of ethical risks, as you must ensure your experiments do not 'damage' your human test subjects, says Reidsma. "You must carefully consider how much impact your research has on your subjects. For example, if you want to train technology to recognise emotions, what videos will you show them to generate anger or sadness? When it comes to interaction technology, you must examine in advance whether the product you are researching is ethically desirable. Furthermore, if you are creating and designing learning technology but you only test it on a very limited target group, then you exclude other groups. This is not acceptable as everybody has the right to benefit from innovation. If you choose to only test sports technology on young and fit 25-year-olds, then it cannot be used by people with major or minor disabilities."
Reidsma adds that the EC will examine whether your research includes any ethical issues that you may have failed to spot and subsequently advises researchers and students on how to improve the proposal. "For example, researchers working for the first time with people who are easily overstimulated often have to be reminded to implement a stop protocol when testing flashy games."
Good knowledge of general principles, less knowledge of specific situations
Reidsma says that most students and researchers have a good knowledge of the general principles and rules, "such as ask permission when involving human subjects in research, pay attention to 'bias' within AI, don't damage systems when conducting 'real world hacking' research and inform the relevant parties if you discover a data leak so they can resolve it. However, they often don't know the finer details of how these general principles apply to their specific research project, for example, if you conduct research involving children, you must ask the parents for permission, and for young people you must ask both the young people and the parents. However, what exactly do you need to ask them? Do you need permission to collect the data, to publish it, or both? Must the human subjects be anonymised or is specific personal data required (for co-design)? The way these and other factors apply to your research is different for each individual project."
For many years, the EC has noticed that researchers and students lack detailed knowledge of the rules in more specific situations as they have never encountered such scenarios in the past. At the same time, the departments and the committee members have formulated a substantial range of best practices, such as tips and tricks for conducting research with people with literacy deficits or with children at a school. Until now, the committee members used to send these tips to the applicants via e-mail, although this took up a lot of time. Reidsma first started to write a handbook, but he quickly realised that this would be a massive document. "If I'd written everything down, it would have taken up hundreds of pages, so after ten pages, I stopped."
Inspired by the Tax and Customs Administration
"We can't expect people to trawl through a 100-page document to find the specific tips and advice applicable to their situation," he grins. "So the challenge was to create something that would quickly provide researchers tips and tricks for creating effective research proposals with very little effort required." Inspired by the online form used by the Tax and Customs Administration, he formulated a number of 'probing questions' for each department and placed them in a decision tree. Your answers to these questions will then guide you to subquestions, which in turn will point you in the direction of the tips and advice relevant to your specific situation.
Think for yourself
"The advice you receive will not be exhaustive," says Reidsma. "And the research proposal will still have to be evaluated by the EC. It is important to us that students and researchers continue to carefully think things through themselves. How we formulate the advice questions is therefore vitally important, meaning a completely different puzzle has to be solved every time. I was assisted in this process by a student in their graduation phase. Ultimately, you may require thousands of advice fragments to provide comprehensive advice to the EEMCS faculty. We expect that if people start using the tool, it will help them to make their research plans more complete and boost the quality. This will mean less work for both the applicants and the committee members."
Help and support from UT colleagues
In order to develop the tool and formulate the advice fragments, Reidsma frequently spoke to a group of around 15 colleagues from the EEMCS departments. He is very grateful for the support he received from his colleagues all over UT while he was developing the tool. Wilma Meere, programme manager of the Digital Competence Centre, provided support with project management and acquisition of funding. LISA (Library, ICT-Services & Archive), the EEMCS faculty and Reidsma's department also contributed financially towards the tool's development. Meere and UB director Marjolein Drent also helped to raise awareness of the tool within other UT faculties and at other universities. The tool was designed in such a way that it allows other faculties and departments to easily fill in the question frameworks and advice fragments. Reidsma also believes the tool will be suitable for other decision processes, such as Research Data Management.
Are you interested in using the tool within your faculty or department? If so, contact Dennis Reidsma.