Learning Technologies & AI: Who are we designing for? Emiliana Murgia1,† , Monica Landoni2,† , Theo Huibers3,† and Maria Soledad Pera4,*,† 1 Università di Genova, Italy 2 Università della Svizzera Italiana, Switzerland 3 University of Twente, The Netherlands 4 Web Information Systems, TU Delft, The Netherlands Abstract Focused on children and the learning context, we argue for the importance of designing artificial intelligence (AI) technologies that take a holistic view of their target users. Rather than prioritize system performance, these smart technologies can be tuned to assist users throughout the task completion process. We argue that considering children’s diverse cultural, social, and emotional backgrounds is critical to pursuing inclusive and adaptive technologies that foster children’s personalised learning and development. Keywords Children, Educational Technology, Personalization, Artificial Intelligence 1. Designing Smart Learning believe these to be crucial when designing technology for education. Moreover, Artificial intelligence (AI), de- Technologies for Inclusion: A fined as “a system’s ability to correctly interpret external Holistic approach data, learn from such data and use those learnings to achieve specific goals and tasks through flexible adap- We live surrounded by technologies that are supposed to tation" [2], could support the development of adaptive help and entertain us. Their development and improve- smart systems. Still, these are generally built upon large ment travel at the speed of light, sometimes delivering on user data sets where individual differences are not visible the promise of improving our life and others failing. For and cannot be catered for specifically. Pandemics and instance, a dyslexic person can overcome difficulties in wars have entered our schools, with more complexity learning thanks to devices and software designed to meet to manage. Can the existing technologies support us? their needs. On the other hand, one of the side issues Could we rely on them to deal with these scenarios and that affect dyslexic people–low self-esteem–is not taken support individual needs? into consideration in the design of the aforementioned Research has proved that promoting high-quality edu- software. Technologies can sometimes lighten or solve cation significantly impacts social and cultural inclusion. some disabilities and relieve us from heavy or risky tasks, Side by side with teachers, technology has become a but can they foster the issues coming from differences in strategic ally in teaching, and learning [3]. Motivated by cultural, social, and emotional backgrounds? User expe- these facts, in this position paper, we argue that if de- rience (UX) research is often centred on user needs. It signers focus their projects on the outcome rather than considers three dimensions: (1) ecological, focused on on supporting the specific tasks and users themselves, the influence of the environment the new technology is whichever they might be, the answer is no. Consider the going to introduce, (2) interaction, describing how the two situations below: users will interact with it and (3) emotional, which aims at making desirable the design [1]. Here, the cultural and • Emma is an 8-year-old clever student. She is active social dimensions are not explicitly accounted for, yet we in interacting during discussions and brainstorm- ing. Teacher Mari introduced Agenda 2030 and Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, asked the class to search for general information Australia about it. Emma is stuck and appears unable to * Corresponding author. complete the task. Since Mari knows that Emma is † Authors contributed equally to this work. timid and needs encouragement and reassurance $ emiliana.murgia@unige.it (E. Murgia); monica.landoni@usi.ch to start, she comes to Emma’s desk and talks to her. (M. Landoni); t.w.c.huibers@utwente.nl (T. Huibers); M.S.Pera@TUDelft.nl (M. S. Pera) That seems to work; Emma is now more confident  0000-0002-9728-1771 (E. Murgia); 0000-0003-1414-6329 and ready to try. (M. Landoni); 0000-0002-9837-8639 (T. Huibers); • Also aged 8, born in Italy to a Peruvian couple, 0000-0002-2008-9204 (M. S. Pera) Nicola is a gentle and intelligent boy. He speaks © 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). Italian at school, and his parents talk to him in CEUR CEUR Workshop Proceedings (CEUR-WS.org) Workshop Proceedings http://ceur-ws.org ISSN 1613-0073 Spanish at home. The result is that he needs help and technologies” [7]. We can imagine a future with with writing texts using the Italian language be- AI designed to avoid the already-mentioned risks. Nev- cause Spanish phonetics causes issues. Teacher Mari ertheless, Nicola and Emma need AI technologies that gives him additional support as soon as the task respond to the child as a whole, not just focused on the requires him to write, as happened when she asked task; an AI that emulates teachers’ approach when sup- the class to search about Agenda 2030. porting students by adapting their actions to context and traits descriptive of individuals. A starting point for this Nowadays, teachers can rely on tools that target a spe- discussion can be grounded on ongoing studies on the cific task: Nicola’s teacher has the support of translators application of AI in education, but extending their reach and platforms that can help international students to as most of these studies focus on remote/online learning learn a new language. On Emma’s side, dyslexia has nu- [8]. Another anchor is the research and design of child merous compensatory and dispensation tools to choose personas in a child co-design process [9] as it allows de- from. In both cases, technology supports learning but signers to target child-specific needs better [10]. In the not enough to let Nicola and Emma feel capable of work- end, we argue for the need for researchers and practition- ing autonomously as their mates. What is missing? The ers in the broad areas of computer science and AI -but teacher has a holistic vision and comprehension of each also colleagues in education and experts in other areas of student; she/he can intervene to ease the learning path study that can support diversity- to continue to allocate whenever needed on different kinds of necessities (techni- efforts to the personalization for inclusion. The aim is cal, social, and others). Research and design approaches to support children’s holistic development and owner- optimised/focused on performance improvement of a ship of their learning while improving how teachers can group of users, even at the cost of personalisation, should support the learning process of their students [11, 12]. adopt and allocate resources to research and develop tech- nologies that have the user and, therefore, personalisa- tion at the centre that should ease the process. To help References teachers, educators, and service workers to be more effi- cient and able to foster personalised support to different [1] R. Hartson, P. Pyla, The UX Book: Agile UX De- people even in less time, we need technologies that can sign for a Quality User Experience, Elsevier Sci- be more like Emma’s teacher. ence, 2018. URL: https://books.google.ch/books?id= Adaptation in the Edtech realm often responds to a Pp5RjgEACAAJ. “single” perspective, e.g., a specific age or skill; also, with [2] A. Kaplan, M. Haenlein, Siri, siri, in my hand: search tools (e.g., visual query formulation for a particu- Who’s the fairest in the land? on the interpreta- lar age group) [4, 5]. The vision for the future is to design tions, illustrations, and implications of artificial in- systems that can adapt to context and traits descriptive of telligence, Business Horizons 62 (2019) 15–25. individuals. Emma and Nicola deserve interactive inter- [3] J. Sparkes, Schools, education and social exclusion, faces and other technologies (algorithms) that can adapt LSE STICERD Research Paper No. CASE029 (1999). and respond on-the-fly to a “holistic" view of the user and [4] J. A. Fails, M. S. Pera, O. Anuyah, C. Kennington, his/her context. They need technologies that consider K. L. Wright, W. Bigirimana, Query formulation cultural, social, emotional, learning, and cognitive fac- assistance for kids: What is available, when to help tors, impacting what children -or adults- need and how & what kids want, in: Proceedings of the 18th ACM they react to technology. AI, along with AI for Education international conference on interaction design and (AIED), embody this vision, even if not yet completely. children, 2019, pp. 109–120. AI technologies are already entering schools and univer- [5] K. Reuter, A. Druin, Bringing together children and sities: intelligent robots and adaptive learning systems, books: An initial descriptive study of children’s for instance, help educators on the path to personalised book searching and selection behavior in a digital learning. It comes with concerns that experts and in- library, Proceedings of the American Society for ternational agencies are underlining regarding incorpo- Information science and Technology 41 (2004) 339– rating AI into the classroom, as this has “the potential 348. to improve education, it may also introduce unforeseen [6] A. Alam, Employing adaptive learning and intel- complications. Artificial intelligence in education will ligent tutoring robots for virtual classrooms and be more effectively utilised in the future by people who smart campuses: Reforming education in the age are aware of the possible drawbacks associated with its of artificial intelligence, in: Advanced Comput- usage” [6]. ing and Intelligent Technologies, Springer, 2022, pp. Tackling the future of education, UNESCO has focused 395–406. on AI in learning environments and the need to consider [7] UNESCO, Recommendation on the ethics the risks and prevent “asymmetric access to knowledge of artificial intelligence, 2021. URL: https://en.unesco.org/about-us/legal-affairs/ A. Soro, M. Brereton, Child designers creating per- recommendation-ethics-artificial-intelligence. sonas to diversify design perspectives and concepts [8] X. Chen, H. Xie, D. Zou, G.-J. Hwang, Ap- for their own technology enhanced library, IDC plication and theory gaps during the rise of ’18, Association for Computing Machinery, 2018. artificial intelligence in education, Comput- [11] D. Schiff, Education for ai, not ai for education: ers and Education: Artificial Intelligence 1 the role of education and ethics in national ai pol- (2020) 100002. URL: https://www.sciencedirect.com/ icy strategies, International Journal of Artificial science/article/pii/S2666920X20300023. doi:https: Intelligence in Education 32 (2022) 527–563. //doi.org/10.1016/j.caeai.2020.100002. [12] E. Murgia, M. Landoni, M. S. Pera, T. Huibers, [9] A. N. Antle, Child-personas: fact or fiction?, in: When will the promises of search technology in Proceedings of the 6th conference on Designing the classroom come true?, in: ICERI2019 Proceed- Interactive systems, 2006, pp. 22–30. ings, IATED, 2019, pp. 10409–10415. [10] H. Itenge-Wheeler, H. Winschiers-Theophilus,