=Paper=
{{Paper
|id=Vol-3762/594
|storemode=property
|title=Towards Trustworthy AI in Inclusive Education: A Co-Creation Approach Rooted in Ecological Frameworks
|pdfUrl=https://ceur-ws.org/Vol-3762/594.pdf
|volume=Vol-3762
|authors=Valeria Cesaroni,Martina Galletti,Eleonora Pasqua,Daniele Nardi
|dblpUrl=https://dblp.org/rec/conf/ital-ia/CesaroniGPN24
}}
==Towards Trustworthy AI in Inclusive Education: A Co-Creation Approach Rooted in Ecological Frameworks==
https://ceur-ws.org/Vol-3762/594.pdf
Towards Trustworthy AI in Inclusive Education: A
Co-Creation Approach Rooted in Ecological Frameworks
Valeria Cesaroni1,* , Martina Galletti2,3 , Eleonora Pasqua3,4 and Daniele Nardi3,5
1
University of Perugia, Italy
2
Sony Computer Science Laboratories-Paris (Sony CSL-Paris) - France
3
Sapienza University of Rome - Italy
4
Centro Ricerca e Cura di Roma - Italy
5
CINI-AIIS - Italy
Abstract
The integration of digital technology and AI systems has prompted extensive inquiries into their ethical design and im-
plementation. Trustworthy AI, essential for fairness, robustness, safety, and transparency, is recognized as fundamental,
transcending pure technological interaction. This is particularly pivotal in contexts involving minors, and especially within
technologies fostering inclusion for students with disabilities and special educational needs (SEND). Inclusive education hinges
on trustworthiness due to inherent asymmetries in learning structures and the relational aspect of educational environments.
This paper starting from a case studies, i.e. an AI interface tailored to children with text comprehension difficulties, it
introduces a co-creation strategy with different stakeholders. It develops an ecological framework for trustworthiness, rooted
in value-sensitive design. Our approach emphasizes ethical and trustworthy AI development, prioritizing responsibility,
reliability, and inclusivity. It addresses the concept of trustworthiness as a systemic relationship between multiple contexts
(such as clinical and school environments) in the development of children’s proximal processes and scaffolding, as outlined
by Bronfenbrenner’s system ecological theory.
Keywords
Trustworthiness, Inclusion, Natural Language Processing, Text Comprehension
1. Introduction most multifaceted and the most interconnected with all
the others domains.
In recent years, the increasingly technologically medi- This is especially true and even more crucial if we
ated nature of our societies, fueled by the systematic think about the use of AI with and by minors and, partic-
integration of digital technology and the escalating ubiq- ularly, within technologies aimed at fostering inclusion
uity of AI systems have prompted increasingly profound in contexts of students with disability and special educa-
inquiries into the ethical, responsible, and trustworthy tional needs (SEND). Within the framework of inclusive
design and implementation of AI. This trend is further education and learning, the aspect of trustworthiness is
underscored by the ongoing legislative endeavors at an pivotal for several reasons:
European and international level. • The inherent asymmetry in the structure of learn-
While there is no universally accepted definition of ing, educational, and rehabilitative interactions.
trustworthy AI, it is widely acknowledged as a fundamen- • The relational aspect of educational and rehabili-
tal principle [1], underpinning the validity of other es- tative environments, grounded in trust and com-
sential characteristics such as fairness, robustness, safety, municative exchanges.
and transparency. Contrary to the notion that trust is • The developmental trajectory of the child, which
solely tied to technology interaction, as suggested by encompasses cognitive and psychological dimen-
Luhmann, trust is deemed indispensable for any form of sions.
interaction to occur [2]. Consequently, in the domain of Considering the intricate nature of the educational and
AI ethics, the concept of trustworthiness is probably the learning dynamics, the conscientious and trustworthy
integration of AI-based technologies within this domain,
Ital-IA 2024: 4th National Conference on Artificial Intelligence, orga- particularly when involving vulnerable subjects, neces-
nized by CINI, May 29-30, 2024, Naples, Italy sitates a thorough evaluative analysis. Such an analysis
*
Corresponding author.
should encompass an examination of the interplay among
$ valeria.cesaroni@dottorandi.unipg.it (V. Cesaroni);
martina.galletti@sony.com (M. Galletti); e.pasqua@crc-balbuzie.it the different systems and players involved.
(E. Pasqua); nardi@diag.uniroma1.it (D. Nardi) In this paper, we start by a case study to illustrate the
0000-0001-9881-1597 (V. Cesaroni); 0009-0002-2079-8999 example of a technology built for therapeutic and ed-
(M. Galletti); 0000-0002-7153-6094 (E. Pasqua); ucational purposes, with a focus on inclusivity, ethical
0000-0001-6606-200X (D. Nardi)
© 2024 Copyright for this paper by its authors. Use permitted under Creative Commons License considerations, and trustworthiness. More, in particular,
Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
we first introduce the ARTIS project, which aims to de- tion, inclusion is addressed primarily through the lens of
velop software using natural language processing (NLP) non-discrimination and data accuracy [11]. While ensur-
tailored to therapeutic practices for children with text ing data quality and mitigating bias are crucial for trust-
comprehension difficulties. Unlike technologies primar- worthy AI systems, this approach risks oversimplifying
ily targeted at disabilities and later repurposed for edu- inclusion as a technical issue, detached from the complex-
cation, ARTIS is conceived as an inherently educational ity of learning environments. Therefore, it’s imperative
tool and subsequently tailored for the clinical world. to broaden the discourse beyond technical considera-
Secondly, we propose a co-creation approach grounded tions and adopt a perspective that encompasses diverse
in value-sensitive design, involving key stakeholders to stakeholders, design aspects, and values promoted in the
ensure the ethical development of the AI interface. Draw- development and implementation process. Despite the
ing from ecological frameworks, such as Bronfenbren- recognized importance of stakeholder involvement, re-
ner’s ecological systems theory [3] , and considering search indicates a lack of participation from clinicians,
that learning environments are both physical and social parents, and teachers in decision-making processes [12],
contexts [4], we emphasize the importance of consid- [13], with limited focus on the broader impact of AI tools
ering the systemic interaction between individuals and on learning environments [14]. This underscores the
their environments in the design process. Our strategy necessity for a more holistic approach to AI education
prioritizes the responsible and trustworthy introduction that considers the multifaceted nature of inclusion and
of AI technology, reflecting our commitment to ethical, engages diverse stakeholders to ensure its effectiveness
reliable, and inclusive development practices. and relevance. This broader perspective aligns with the
The issue of AI trustworthiness is therefore positioned framework proposed by [9], which emphasizes inclusive
at the crossroads of various interaction systems known as AI education grounded in Universal Design for Learning
proximal processes. These processes are pivotal because principles.
they serve as the conduit through which individuals learn The integration of Artificial Intelligence holds promise
and adapt over time within their interactional environ- for children with special learning needs, including those
ments [5]. with visual, mobility, or hearing impairments, offering
opportunities for diverse learning materials and adapt-
able content strategies [15]. However, achieving true
2. AI and Inclusion inclusion extends beyond technological solutions alone.
Literature emphasizes the importance of striking a bal-
The ARTIS project presented in this paper positions itself
ance between individualization and socialization, valuing
in the debate on the relationship between AI and inclu-
differences while addressing diverse needs [16]. There-
sion in an innovative way. It develops a technology that
fore, inclusive contexts must embrace both technological
assists children with special educational needs and learn-
advancements and nuanced pedagogical approaches to ef-
ing disorders, thus aligning itself with the field of assis-
fectively support diverse learners. This perspective aligns
tive technologies. Yet, it doesn’t stop at the deterministic
with the holistic framework proposed earlier, which ad-
association that sees inclusion as directly conveyed by
vocates for inclusive AI education grounded in Universal
technology, but problematises this relationship through
Design for Learning principles and comprehensive stake-
a systemic vision of inclusion and through an approach
holder engagement. It It becomes crucial, therefore, to
based on the ethics of AI and the active involvement
reflect on how the relationship between technology, dis-
of numerous stakeholders in all phases of the project.
ability, and inclusion should be conceptualized.
While there is extensive literature on AI and inclusion,
The concept of inclusion, related to that of disability,
few studies explore the intersection of these fields, such
has changed significantly over time as the work of [17]
as those by Kazimzade et Al. [6] and Gibellini et Al. [7].
and [18] testifies. In 1980, the International Classification
Similarly, in the domain of AI ethics in education and
of Impairments, Disabilities, and Handicaps [19] estab-
learning, research is limited, with Mouta et Al. [8] being
lished the first definition of disability, which is commonly
among the few contributors. The only paper addressing
associated with a medical model of disability. Accord-
this gap is Song et Al.’s recent study [9] introduces a
ing to this view, disability is interpreted as an individual
framework grounded in Universal Design for Learning
problem, which must be corrected in terms of restitutio
principles [10], guiding the development of inclusive AI
ad integrum. In this publication, an important distinction
education. By incorporating essential AI concepts and
was made between impairment (an anomaly or loss of
diverse pedagogical examples, this framework aims to
physiological or anatomical functions), disability (the loss
foster broader participation and bolster competitiveness
or difficulty, resulting from the impairment, of perform-
within the AI workforce.
ing activities considered normal for healthy individuals),
In the context of European and international guide-
and handicap (the condition of disadvantage resulting
lines, as outlined by UNESCO specifically for AI in educa-
from the objectification of the impairment). According
to this conception, disability is seen as something exclu- renders it a critical resource for enhancing people’s lives
sively pertaining to the individual and limits her ability to and expanding their opportunities. However, it is essen-
live a full and satisfying life and to flourish in society [20]. tial to recognize its situated nature, shaped by social and
Disability movement activists in the late 20th century value determinants. Technology thus becomes a matter
posed a significant challenge to this prevailing paradigm. of justice. Consider, within the realm of education, the
They critically examined the connection between impair- extent to which knowledge can be made accessible to
ment and disability, thus affirming the social model of diverse needs and characteristics through design—both
disability [21]. This conception emphasises the role of the technological and otherwise—that is attentive to differ-
material organisation of society in creating conditions ences, as articulated in Universal Design for Learning
of disability from a deficit. Therefore, in this view an (UDL).
inclusive society is one that can eliminate the disabling These considerations are particularly pertinent regard-
contexts and barriers that society possesses in relation ing AI-based systems. Fourth-order technologies [28],
to the diversity of human functioning. One of the most capable of replicating human actions in the absence of hu-
significant steps in this debate was the development of man intelligence, bring with them potentials, challenges,
the ICF model [22]. At the centre of the bio-psycho-social and risks well-documented in literature and institutional
approach, there is the individual-environment interac- discourse. For inclusive learning objectives, AI-based
tion. Disabilities are considered beyond an exceptionalist technologies must be designed and implemented with
logic and are placed at the centre of the human condition adherence to various criteria to ensure accountability
itself. What links disability and health, according to this and reliability within an inclusive learning environment.
model, depends on the functionings or, in other words, Beginning with an assessment of the ethical treatment
the capabilities that each human being in relation to the of data (bias, privacy) and algorithms (fairness, trans-
social context of reference is capable of having and being. parency, explainability), an inclusive perspective on tech-
Thus, the ICF model, and the Capability Approach (CA), nology prompts a critical examination of its embedding
[23],[24] although not totally overlapping, emphasise context and its interaction with and involvement of key
how well-being (individual and collective) arises from stakeholders. Moreover, from a technical perspective,
the dynamic between internal dispositions/opportunities technologies for inclusion should be designed so that the
and external dispositions/powers, a dynamic that is to be users can personalize their interaction with the AI system
understood both as the absence of external impediments through adjusting parameters and controls according to
(negative freedom) and as an analysis of the conditions their preferences or specific needs.
and quality of life that people are able to lead (positive
freedom).
In this context, the role of technology acquires a cen- 3. ARTIS: system description
tral function, in fact, the use of technology in relation
In designing AI-based technologies for inclusive learn-
to disability could play a crucial role in the expansion
ing objectives, such as the interface ARTIS developed by
of individuals’ capabilities [25]. Technology finds a spe-
Sony Computer Science Laboratories Paris and Centro
cific place within the Environmental Factors of the bio-
Ricerca e Cura, ensuring accountability and reliability
psycho-social perspective of the ICF. Starting from the
within an inclusive learning environment is paramount.
assumption that people can function in different ways
This necessitates adherence to various criteria, beginning
depending on their environments and that disability is
with an ethical assessment of data and algorithms, encom-
therefore the result of a mode of interaction between
passing considerations of bias, privacy, fairness, trans-
the individual and the environment, the ICF considers
parency, and explainability. An inclusive perspective on
technologies as tools that mediate this interaction. Tech-
technology calls for a critical examination of its embed-
nologies thus, depending on the way they are designed,
ding context and its interaction with key stakeholders.
implemented, and used, can act as facilitators or, on the
ARTIS, as an interface powered by artificial intelligence
contrary, as barriers in performing normal activities and
to support text comprehension, exemplifies this approach
creating an inclusive society. It would be naive to con-
by drawing on neuro-psycholinguistic models of reading
sider technology’s relationship with disability solely in
comprehension, thus integrating linguistic components
instrumental terms. As extensively debated in the lit-
into its design to enhance accessibility and inclusivity in
erature ([26], [27]), technologies always exist within a
learning contexts. ARTIS [29] is an interface, powered
specific context and invariably embody certain underly-
by artificial intelligence, designed to support text com-
ing values and orientations. In the context of disability,
prehension. Born as a collaboration between Sony Com-
technology assumes a pivotal role, often becoming an
puter Science Laboratories Paris 1 and Centro Ricerca
integral part of an individual’s interaction with their
environment. Technology’s capacity to mediate this re-
lationship between the individual and the environment 1
https://csl.sony.fr/
e Cura 2 in Rome, Italy. The interface was developed 4. Ecological framework for a
from neuro-psycholinguistic models of reading compre-
hension, focusing on the linguistic components of text
trustworthy AI with children
processing. The project description underlines how the objectives of
In particular, subjects with poor text comprehension this project are multifaceted. It emphasize rigorous re-
present difficulties related to the processing of syntactic search to ensure computational functionality, facilitating
and semantic sentence components [30], the analysis of integration into rehabilitation and educational activities.
lexical components of words [31] and deficits in the syn- Additionally, specialized research is essential to examine
tactic representation of words and oral comprehension the impact, acceptance and feasibility within inclusive
skills [32]. Moreover, [33] stated that the same subjects educational and rehabilitative contexts. To address this,
report significant deficits in receptive vocabulary and se- we framed future research directions in a conceptual
mantic processing. Finally, [34] and [35] addressed the framework considering impacts on child development
issue of grammar, claiming that children and adolescents and the interplay between inter-subjective and objectual
with problems in text comprehension show difficulties dimensions of proximal development.
in understanding the role of pronouns within sentences, The literature on AI trustworthiness is vast and mul-
especially if these are in clitic form. Considering this ap- tidisciplinary. With regard to HCI (human computer
proach, ARTIS allows for personalized practice on texts interaction), the issues that are most focused on concern
at different levels. Using AI algorithms, the interface can the design and perception of the user and the psycho-
automatically extract keywords, associate pictograms, logical mechanisms that impact the perception of trust-
identify more complex vocabulary and generate semantic worthiness and thus the subsequent usage behaviour
networks, and practice grammatical components. ARTIS [36],[37],[38]. Therefore, one of the main focuses is on
is aimed at primary and secondary school children with predictability, transparency, explainability, robustness of
difficulties in text comprehension, but it can also be used the system. Trust is generally interpreted as a psycholog-
as a support for L2. ical mechanism that occurs in social or intersubjective
We had three main goals in mind when designing interaction to reduce the uncertainty of the other’s be-
our prototype. Firstly, we aimed to develop a system haviour.
specifically for children from second grade to adoles- According to the OECD, Organisation for Economic
cents diagnosed with reading comprehension difficulties. Co-operation and Development : “AI might be considered
Secondly, we wanted an interface suitable for therapeu- trustworthy when it does properly what it is supposed to do,
tic sessions, always under professional supervision for but also when one can trust that human beings will use it
younger age groups. Lastly, we aimed to involve end- in a fair and appropriate way”[39]. The trustworthiness
users and stakeholders from the outset of the design of AI is also considered a crucial element within the edu-
process, collaborating with skilled speech and language cational literature, and involves the aspects of reliability,
therapists for evaluation and feedback. Drawing inspi- fairness, transparency, explainability, data protection and
ration from Kitsch and Van Dijk’s model, we integrated bias.
functionalities to enhance the superficial representation However, we believe that the context of the ARTIS
of language, focusing on lexical and morphosyntactic project poses an even greater challenge to the concept
understanding. Our interface consists of three modules: of trustworthy AI, because it must be ensured in such a
the first aids in understanding words and sentences, the way that this characteristic ’collapses’ within different
second focuses on coherent sequence representations, settings and interactions: rehabilitation, school, family
and the third aims at creating a broader mental model of and individual use. The interface, in fact, not only hy-
language. We also implemented a Synset Networks fea- bridises the educational and rehabilitative relationship
ture to link words encountered in the text with previous in a consistent manner, but also assumes a scaffolding
knowledge. This approach fosters vocabulary expansion function for the child, crucial for proximal development
and deeper understanding of word meanings. Overall, processes: that is, the difference between an individual’s
our design reflects a comprehensive approach to address actual level of development and the level that can be
reading comprehension challenges reached, through the help of a tutor or through social
interaction [40]. Therefore, the ARTIS project shows
how it is necessary to take a holistic view in order to
correctly interpret the issue of trustworthiness within an
inclusive learning context. In particular, the ecological
system theory elaborated by Bronfenbrenner turns out
to be a strategic framework.
According to the author of the ecological theory of
2
https://www.crc-balbuzie.it/
human development, there are two forms of proximal to trustworthy in inclusive education and speech and
development processes: a) those with other persons and language therapy through the cooperation with a broad
b) those with symbols and objects; and these processes number of stakeholders. Moreover, we outlined the
occur in a dynamic of reciprocal influence and determi- general framework to carry on future research on the
nation of several environmental levels: the microsystem ARTIS project rooted in an developmental ecological
(in which interpersonal relationships occur directly and theory and cooperation strategies. Future work in this
in defined contexts, such as family, school, peers), the area should focus on developing an integrated approach
mesosystem (relationships between microsystems, for to assess the impact and acceptance of technology in
example), the macrosystem (cultural and social norms), educational settings, with a particular emphasis on
the exosystem (environments that influence indirectly, engaging a more diverse range stakeholders such as
such as parental work) and the chronosystem (temporal teachers, children and families. Building upon existing
changes, such as historical events and life transitions). research, future studies could explore innovative
However, the nature of digital and AI technologies methodologies that combine quantitative metrics with
is precisely to hybridise these systems and their inter- qualitative insights to gain a holistic understanding of
actions [41]. The proximal process therefore occurs in the complex dynamics involved. Additionally, there is a
a hybridised manner, involving both interpersonal re- need to investigate the long-term effects of technology
lations and object relations, and collapsing the interac- integration on learning outcomes and socio-emotional
tions between different levels even more directly, There- development across various age groups. Collaborative
fore, it is crucial to ask, for example, in ARTIS how does efforts involving researchers, educators, policymakers,
the macrosystem and exosystem (the decisions made by and technology developers will be essential to address
the developers) affect the interactions in the microsys- the multifaceted concerns raised by teachers, families,
tem? How does the hybridisation of the microsystem therapists, and children, thereby fostering a more
(class, family, rehabilitation) reshape the concepts and inclusive and supportive educational environment.
processes of inclusion? It’s crucial to consider not only
the impact of technology failures, such as malfunctioning We acknowledge partial financial support from
algorithms or discriminatory biases, but also its intended PNRR MUR project PE0000013-FAIR
outcomes. To address these questions comprehensively,
a collaborative strategy engaging all educational stake-
holders—teachers, clinicians, parents, and students—is References
essential for qualitative and quantitative insights.
[1] H. AI, High-level expert group on artificial intelli-
gence, Ethics guidelines for trustworthy AI 6 (2019).
5. Cooperation Strategies [2] N. Luhmann, Trust and power, John Wiley & Sons,
2018.
As the project interfaces blend educational and rehabilita- [3] U. Bronfenbrenner, The ecology of human develop-
tive dynamics, it becomes clear that a holistic approach is ment: Experiments by nature and design, Harvard
indispensable. In ARTIS, a collaborative strategy is essen- university press, 1979.
tial, involving stakeholders across education—teachers, [4] C. Ingleton, Emotion in learning: a neglected dy-
clinicians, parents, and students—to ensure AI trustwor- namic, in: HERDSA annual international confer-
thiness in inclusive learning environments. This mul- ence, Melbourne, Citeseer, 1999, pp. 12–15.
tifaceted cooperation strategy integrates academic and [5] L. S. Vygotsky, Thought and language, MIT press,
industry channels, with early involvement of speech and 2012.
language therapists ensuring interdisciplinary engage- [6] G. Kazimzade, Y. Patzer, N. Pinkwart, Artificial in-
ment. After the initial proof of concept, the interface telligence in education meets inclusive educational
underwent testing by children, therapists, and the public, technology—the technical state-of-the-art and pos-
promoting User-Centered Design principles and enabling sible directions, Artificial intelligence and inclusive
ongoing monitoring. Focus groups involving developers, education: Speculative futures and emerging prac-
therapists, and ethics experts ensured ethical oversight, tices (2019) 61–73.
fostering critical thinking and responsible technology [7] G. Gibellini, V. Fabretti, G. Schiavo, Ai education
usage. from the educator’s perspective: Best practices for
an inclusive ai curriculum for middle school, in:
Extended Abstracts of the 2023 CHI Conference on
6. Conclusion & Future Work Human Factors in Computing Systems, 2023, pp.
1–6.
In this paper, we presented the ARTIS project as a
[8] A. Mouta, A. M. Pinto-Llorente, E. M. Torrecilla-
significant case study illustrating a systemic approach
Sánchez, Uncovering blind spots in education [25] N. Coombs, Disability and technology: A historical
ethics: Insights from a systematic literature review and social perspective. (1990).
on artificial intelligence in education, International [26] B. Friedman, Value-sensitive design, interactions 3
Journal of Artificial Intelligence in Education (2023) (1996) 16–23.
1–40. [27] D. J. Gunkel, The machine question: Critical per-
[9] Y. Song, L. R. Weisberg, S. Zhang, X. Tian, K. E. spectives on AI, robots, and ethics, mit Press, 2012.
Boyer, M. Israel, A framework for inclusive ai [28] L. Floridi, The fourth revolution: How the info-
learning design for diverse learners, Computers sphere is reshaping human reality, OUP Oxford,
and Education: Artificial Intelligence (2024) 100212. 2014.
[10] C. CAST, Universal design for learning guidelines [29] M. Galletti, E. Pasqua, F. Bianchi, M. Calanca,
version 2.0, Author Wakefield, MA (2011). F. Padovani, D. Nardi, D. Tomaiuoli, A reading
[11] W. Holmes, F. Miao, et al., Guidance for generative comprehension interface for students with learning
AI in education and research, UNESCO Publishing, disorders, in: Companion Publication of the 25th In-
2023. ternational Conference on Multimodal Interaction,
[12] M. Rowe, Shaping our algorithms before they shape 2023, pp. 282–287.
us, Artificial intelligence and inclusive education: [30] J. Oakhill, Instantiation in skilled and less skilled
Speculative futures and emerging practices (2019) comprehenders, The Quarterly Journal of Experi-
151–163. mental Psychology Section A 35 (1983) 441–450.
[13] P. Kousa, H. Niemi, Artificial intelligence ethics [31] K. Nation, M. J. Snowling, Factors influencing
from the perspective of educational technology syntactic awareness skills in normal readers and
companies and schools, Learning: Designing the poor comprehenders, Applied psycholinguistics 21
Future (2023) 283. (2000) 229–241.
[14] W. Holmes, I. Tuomi, State of the art and practice [32] R. Padovani, La comprensione del testo scritto in
in ai in education, European Journal of Education età scolare. una rassegna sullo sviluppo normale e
57 (2022) 542–570. atipico, Psicologia clinica dello sviluppo 10 (2006)
[15] L. Cottini, et al., Didattica speciale e inclusione sco- 369–398.
lastica, Carocci, 2017. [33] K. Nation, M. J. Snowling, Beyond phonological
[16] A. Lascioli, et al., Verso l’inclusive education, Edi- skills: Broader language skills contribute to the
zioni del Rosone, 2014. development of reading, Journal of research in
[17] S. Vehmas, Ethical analysis of the concept of dis- reading 27 (2004) 342–356.
ability, Mental retardation 42 (2004) 209–222. [34] F. Arosio, C. Branchini, L. Barbieri, M. T. Guasti,
[18] L. Terzi, A capability perspective on impairment, Failure to produce direct object clitic pronouns as a
disability and special needs: Towards social justice clinical marker of sli in school-aged italian speaking
in education, Theory and research in education 3 children, Clinical linguistics & phonetics 28 (2014)
(2005) 197–223. 639–663.
[19] W. H. Organization, N. C. for Health Statistics (US), [35] A. Zachou, E. Partesana, E. Tenca, M. T. Guasti,
The International Classification of Diseases, 9th et al., Production and comprehension of direct
Revision, Clinical Modification: Procedures: tabular object clitics and definite articles by italian children
list and alphabetic index, volume 3, Commission on with developmental dyslexia, Advances in language
Professional and Hospital Activities., 1980. acquisition (2013) 464–471.
[20] H. Kuhse, P. Singer, P. Singer, Should the baby live?: [36] R. Lukyanenko, W. Maass, V. C. Storey, Trust in arti-
The problem of handicapped infants, volume 228, ficial intelligence: From a foundational trust frame-
Oxford University Press Oxford, 1985. work to emerging research opportunities, Elec-
[21] M. Oliver, M. Oliver, The politics of disable- tronic Markets 32 (2022) 1993–2020.
ment—new social movements, The politics of dis- [37] L. P. Robert Jr, G. Bansal, N. Melville, T. Stafford,
ablement (1990) 112–131. Introduction to the special issue on ai fairness, trust,
[22] W. H. Organization, International Classification of and ethics, AIS Transactions on Human-Computer
Functioning, Disability, and Health: Children & Interaction 12 (2020) 172–178.
Youth Version: ICF-CY., World Health Organization, [38] H. Choung, P. David, A. Ross, Trust in ai and its role
2007. in the acceptance of ai technologies, International
[23] S. M. McMurrin, The Tanner lectures on human Journal of Human–Computer Interaction 39 (2023)
values, volume 8, Cambridge University Press, 2011. 1727–1739.
[24] M. C. Nussbaum, Frontiers of justice: Disability, [39] S. Vincent-Lancrin, R. Van der Vlies, Trustworthy
nationality, species membership, in: Frontiers of artificial intelligence (ai) in education: Promises
Justice, Harvard University Press, 2007. and challenges (2020).
[40] N. Yelland, J. Masters, Rethinking scaffolding in the
information age, Computers & Education 48 (2007)
362–382.
[41] J. L. Navarro, J. R. Tudge, Technologizing bronfen-
brenner: neo-ecological theory, Current Psychol-
ogy 42 (2023) 19338–19354.