AIXPERIMENTATIONLAB — Participatory Approach of Graphical User Interface Design
Towards an institutionalized format for the development, use, and diffusion of human-centric artificial intelligence applications
Physical Interface Between Humans and AI
The AIXPERIMENTATIONLAB research project aims to develop human-centered AI applications in the work context. The goal is to reduce employee stress while increasing operational efficiency. To achieve this, the project follows a participatory approach, which involves employees in the design development process of a graphical user interface (GUI). Following this approach, we are addressing the question: how can we design a human-AI interface in a way that users get the most out of it?
The participatory approach is based on the assumption that users are more likely to use and be satisfied with a system if they participate in its development, rather than if it is imposed on them in a finished form. Following this idea, employees from two participating companies were involved in this development. The goal of one use case, for example, is to design an AI that supports employees in the service area when processing customer queries.
Figure 1 shows the process and methods that we used in the human-centered design approach.
1. Information Event
In the begin of the process, the employees had to be enabled to play an active role as designers of their future GUI. Therefore, we began with a remote information session which included basic knowledge about interfaces, respectively a GUI. Furthermore, the participants got to know a variety of design elements to open their minds for the creation of individual ideas. Finally, the technical solution including the input and output parameters was presented in a simplified manner, so that the employees could directly think about how they would like to design those parameters.
2. GUI Workshop
Afterwards, a GUI-workshop took place on-site. For each company, we formed two groups of three to six employees to create at least two different versions of mock-ups. The workshop consisted of a brainstorming session, an individual sketching part and a collaborative sketching part.
First, the employees thought about which information and functions would be essential to integrate in the GUI and what design principles are important for their specific use case. They were guided to create an overview of which features they expect from the user interface. Subsequently, each workshop participant sketched their personal vision of the future GUI on paper. A selection of templates helped them jump-start this task without already imposing design aspects on them. Some of the results are presented in Figure 2. This procedure was chosen to create a variety of sketches without the group members being influenced by the ideas of their co-workers. Each participant presented their own idea to the group and received feedback. The group analysed each proposal. The goal was to reconsider individual functions and presentation methods and to identify strengths and weaknesses. This evaluation finally served as the basis for the joint development of a final version in each group.
3. Mock-Up Creation
The paper-based results of the GUI workshops were transferred into digital mock-ups and the variation of different proposed design elements were elaborated by the project team. This procedure helped us to gain an understanding of the employees’ requirements. Based on them, we build visual blueprints of which features the system should include to best support employees. We adopted functions, information, positions, colours, sizes, and other design aspects from the employees’ versions.
4. Questionnaire
To allow employees to actively participate in deciding on one of the design variations for a mock-up, an online survey was conducted. The survey was spread among the employees. In some companies the sample also included employees who were not able to participate in the workshops, which led to an even greater involvement. The last step of the design process was the evaluation of the survey and compilation of the employees’ favoured design elements into the final mock up.
5. Iterations
The companies realized the frontends inhouse. Nevertheless, due to the iterative process of the development, this first realization will not be the final GUI. Especially when the users start to interact with the AI-solution via the self-designed GUI they will — certainly — require changes and adaptions. In conclusion, the participatory approach had a major share in the development of a realistic expectation of the users towards the solution. Read more about the challenge of expectation management in AI-implementation projects in our paper (Buschmeyer et al., 2022)
Developing AI Application in Social Partnership
The Federal Ministry of Labour and Social Affairs (BMAS) is working on many levels to make developments and the use of AI in the work context transparent and to shape the digital future in a human-centred way. This purpose is addressed, among others, by the Initiative New Quality of Work (INQA, for its acronym in German).
In the field of action “Learning and Experimental Spaces AI” (Lern- und Experimentierräume KI in German), eleven projects are taking up this challenge. They develop strategies for the introduction of non-discriminatory AI in companies in a social partnership, test solutions for further training, and develop company design guidelines to promote digital transformation and the implementation of AI solutions particularly in SMEs across various sectors. All projects involve several partners — from academia and business, from both the employee and employer side — to develop jointly supported solutions. AIXPERIMENTATIONLAB is one of the individual projects funded by BMAS through INQA.
Yearly public events are planned to take place at the AIXLAB to demonstrate the project progress as soon as functional prototypes are developed based on project schedule. For updates and current information please check the project website.
Sources
INQA: INQA-Lern- und Experimentierräume KI. URL: https://inqa.de/DE/initiative-und-partner/projektfoerderung/inqa-lern-und-experimentierraeume-ki/uebersicht.html
Buschmeyer, K., Hatfield, S., Heine, I., Jahn, S., & Markus, A. L. (2022). Expectation management in AI implementation projects: a case study. EuroMed Journal of Business, (ahead-of-print).
