End-user development for democratising artificial intelligence

End-user development (EUD) is a research area that rapidly developed after the European Network of Excellence (EUD-Net) established in 2003. It aims at proposing approaches to empower end users to develop and adapt systems at a level of complexity that is adequate to their expertise, practices, and skills. EUD may occur along the entire software lifecycle, with the purpose of making users able to participate in their artefact development both at design time and at use time.

A first definition of EUD, coined within the EUD-network, was given in a first book on EUD: ‘a set of methods, techniques, and tools that allow users of software systems, who are acting as non-professional software developers, at some point to create, modify, or extend a software artifact’ (Lieberman, Paternò, and Wulf 2006).

Coherently with that definition, originally, scholars proposed methods, techniques and tools that allow end users to modify or extend software artefacts, such as spreadsheets, web applications, video games, and mobile applications. In the so-called Internet of Things era, EUD moved on to address the problem of defining and modifying the behaviour of smart environments, including smart objects, pervasive displays, smart homes, smart cities, and so on (Paternò and Wulf, 2017). Therefore, the term ‘end-user development’ acquired a broader meaning covering approaches, frameworks and socio-technical environments that allow end users to shape digital artefacts that encompass both software and hardware technology (Barricelli et al. 2019).

Recent research and technological trends concerning Artificial Intelligence (AI) approaches and algorithms have contributed to renew the vision of end-user development, by providing tools and platforms that allow end users to harness the power of AI to create solutions that encompass computer vision, image processing, natural language interfaces, as well as easier ways to control smart environments. Similarly, AI bears the promise of facilitating EUD activities by end users, complementing them as ‘co-designer’.

This special issue focuses on these new research trends and addresses the theme of EUD for AI-based systems, considering the current importance of allowing end users to personally control and adapt the behaviour of AI-infused systems, such as intelligent agents, expert systems, autonomous and collaborative robots, recommender systems, and the like. The special issue aims to propose the adoption and experimentation of EUD in new AI-based application fields, and to discuss the impact of EUD on AI-based systems in terms of their acceptability and appropriation by end users.

The four papers in this special issue are the revised versions of the respective papers presented at the International Conference on End-User Development (IS-EUD 2021) (Fogli et al. 2021). They have been initially selected among all those published in the conference proceedings and after a rigorous double-blind review process have been included in the special issue.

The paper by A. Sanctorum, J. Riggio, J. Maushagen, S. Sepehri, E. Arnesdotter, M. Delagrange, J. De Kock, T. Vanhaecke, C. Debruyne, and O. De Troyer presents a EUD approach to democratising the creation of knowledge bases. Knowledge bases that collect information about a specific domain can be used for AI-based data processing; however, their construction usually requires expertise not only in the subject matter but also in the semantic technologies that guarantee their well-structuredness and machine-readability. The EUD approach presented in the paper empowers domain experts in defining concepts and relationships to represent the available knowledge formally, and to set up the knowledge base and fill it with data. The toxicology domain is considered as use case.

R. Andersen, A. I. Mørch and K. T. Litherland investigate the role played by Human-Centered Artificial Intelligence (HCAI) in Education. They explore the use of HCAI for visual artefact analysis and automated scaffolding by an AI-enabled chatbot simulating a teacher’s behaviour in a learning makerspace. Knowledge-based rules have been derived and experimented to support block-based programming of smart objects by pupils on one side and visual analysis of developed artefacts by educational researchers on the other side. In synthesis, the paper illustrates how to obtain human Intelligence Augmentation (IA) through HCAI.

M. Manca, F. Paternò and C. Santoro explore the potential of end-user personalisation in the Industry 4.0 context, where the exploitation of Internet of Things (IoT) combined with AI algorithms is gaining momentum. The study explores the needs and opportunities for EUD within the context of a paper mill, and then proceeds by evaluating a prototype EUD environment in this context. In particular, the authors report on the opportunity for providing workers unskilled in programming with the ability to customise the control and monitoring of Industry 4.0 smart factories to accommodate changing circumstances and specific contexts.

F. Corno, L. De Russis and A. Monge Roffarello conducted a week-long diary study to investigate trigger-action rule specification by end users to personalise their IoT ecosystems. The paper concludes by suggesting design guidelines and opportunities for EUD in this domain. An important consideration resulting from the paper is the need to facilitate different perspectives (device-centric, information-centric, and people-centric) in specifying trigger-action rules. The complexity of adapting interfaces and abstractions to contextual factors is an interesting opportunity for the design of AI-supported interfaces for trigger-action programming.

Acknowledgements

We are grateful to the Editor-in-Chief, Professor Panos Markopoulos, for the opportunity to assemble this special issue. We thank all the authors for their contribution to the special issue and the reviewers for their help in improving the final versions of the papers.