Techville’s Chronicles: A Music Pedagogy Project to Foster Children’s AI

Literacy Through Co-Creativity and Multimedia Storytelling

Emilia Parada-Cabaleiro

Department of Music Pedagogy, Nuremberg University of Music, Germany

Keywords:

Lifelong Learning, 21st Century Skills, STEAM, AI Ethics, Co-Creative Composition, Music Emotions.

Abstract:

The rapid integration of Artiﬁcial Intelligence (AI) in our daily lives has raised an always increasing attention

for the need of developing AI literacy. In the realm of STEM (Science, Technology, Engineering, Mathemat-

ics), successful initiatives aiming to develop children’s AI literacy can be found in the literature. However,

despite STEAM-based methods offer a great potential for music education too, the integration of the (A)rts in

STEM education is mostly biased—the arts are used as a tool to support STEM subjects but not the other way

around. With this background, the question how AI literacy might be promoted in the music classroom through

audiovisual co-creativity and digital storytelling is explored in a music-pedagogy workshop with nine children

(7 male, 2 female) aged 10 to 12. This work is a proof of concept that illustrates how generative tools might be

used to support child-AI co-creative interactions and by this acquiring not only digital skills and knowledge

but musical ones, too; thus, fostering an integrative development of artistic and technical competences.

1 INTRODUCTION

“So-called digital natives, born already immersed in

a digital broth culture, have now internalized the dig-

ital “gesture”, but not the rules and the awareness of

the logic inherent in the technologies they use” (Stri-

ano, 2019, p. 84). The words by Francesco Striano in-

vite to reﬂect about an eventual mismatch between the

myriad of opportunities offered by using Artiﬁcial In-

telligence (AI) in education (Zhang and Aslan, 2021)

and students’ readiness to fully proﬁt from them. An

important requirement to successfully interact with

technology is Digital literacy, a set of competences

including not only the ability to use but also to under-

stand and critically assess the implications of using it.

Digital literacy makes the difference between passive

consumers and sovereign individuals whose sense of

agency proactively shapes the digital landscape.

AI literacy can be considered within the concept

of Digital literacy, which is one of the eight key

lifelong learning competences amongst those recom-

mended by the Council of the European Union as es-

sential “for personal fulﬁlment, a healthy and sustain-

able lifestyle, employability, active citizenship and so-

cial inclusion” (European Commission, 2019, p. 4).

In its last update, the the DigComp framework (Fer-

rari and Punie, 2013), includes an Annex about Cit-

https://orcid.org/0000-0003-1843-3632

izens interacting with AI systems (Vuorikari et al.,

2022, p. 77). Therefore, introducing AI technology in

the classroom from early age seems to be a necessary

step not only to enhance learning but also to develop

an authentic knowledge about AI (Su and Yang, 2022,

p. 4) and therefore promoting Digital and AI literacy.

With this background, a workshop to promote AI

literacy within the music classroom, is presented. A

total of 9 children, aged 10 to 12, participated in the

activity, whose goal was to co-creatively (i. e., by in-

teracting with generative AI) compose the music and

create the illustrations for a multimedia Fairy-Tale

deﬁning AI and narrating its ethical implications. An

English version of the Fairy-Tale is freely available.

The article proceeds as follows. In Section 2,

previous works are outlined. Section 3 presents the

methodology. Sections 4 and 5 describe and discuss

the results. Finally, Section 6 concludes the article.

2 STATE OF THE ART

Recent research highlights the beneﬁts of a success-

ful integration of computational thinking within the

music curriculum (Bell and Bell, 2018) as well as

the great potential of using STEAM-based tools for

https://zenodo.org/records/10825484

Parada-Cabaleiro, E.

Techville’s Chronicles: A Music Pedagogy Project to Foster Children’s AI Literacy Through Co-Creativity and Multimedia Storytelling.

DOI: 10.5220/0012731000003693

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 16th International Conference on Computer Supported Education (CSEDU 2024) - Volume 1, pages 623-630

ISBN: 978-989-758-697-2; ISSN: 2184-5026

623

music education (

Ozer and Demirbatir, 2023). Con-

cerning the development of children’s AI knowledge,

very promising initiatives such as the PopBots toolkit

(Williams et al., 2019) exist. Still, such initiatives,

although integrating music-related activities, are typi-

cally implemented within informatics-related courses.

This relates to the fact that the integration of the Arts

in STEAM is “often misunderstood as referring to the

use of the arts only to enhance teaching and learning

[i.e., arts are used as] a vehicle for learning STEM—

not the other way around” (Liao, 2016, p. 45).

STEAM education should be, however under-

stood as an integration between ‘The Arts’ and ‘The

Science’ (Jesionkowska et al., 2020). The idea behind

this formulation is that solving nowadays problems

requires innovative solutions for which both techni-

cal skills and creative thinking are essential. Such so-

lutions will only emerge from connections amongst

different knowledge domains (Jesionkowska et al.,

2020). From this perspective, educational initiatives

within STEAM often involve the integration between

children’s literature, art, and technology in order to

foster children’s creativity and technological thinking,

e. g., through digital story writing (Ng et al., 2022).

This type of research can be broadly included

within the realm of Computational Creativity, which

centres around three main areas (Tatar and Pasquier,

2019): (i) Artiﬁcial creative systems, typically inves-

tigated within Computer Science; (ii) Computational

systems for supporting creativity, typically explored

by artists; (iii) Computational models of (human) cre-

ativity, often research within psychology which aims

to better understand human creativity by employing

computational models. In this work, we will refer

to the second type of application, i. e., computational

systems, used within an educational setting, to foster

children’s creativity while acquiring AI literacy com-

petences. This will be possible through digital sto-

rytelling, which besides promoting creative attitudes

through digital possibilities has shown to be an opti-

mal way to encourage children discussions about the

impact of AI in our society (Ng et al., 2022).

Depending on the level of involvement played by

the AI, the interaction between the user and a compu-

tational system supporting creativity could be concep-

tualised as co-creativity. As indicated by Kantosalo et

al., co-creative systems “represent a middle ground

between autonomous creative systems, which are in-

tended as the sole shepherds of their own creativity,

and creativity support systems, which instead facil-

itate the creativity of their users” (Kantosalo et al.,

2020, p. 57). Still, since disentangling to which extent

a user’s creativity has been inﬂuenced or activated in

any way by a system might not be totally possible,

Photos: Emine Yaprak Kotzian

Figure 1: Four children and one instructor interacting with

AI tools during the workshop [rights of all photos granted].

a clear distinction between co-creative and creativity

support systems would be difﬁcult in some scenarios.

Beyond this conceptual fuzziness, AI systems

aiming to enhance musicians’ creativity have been

presented. Still, the overwhelming amount of gen-

erated content and the limited interaction possibilities

are often drawbacks preventing a fruitful use of gener-

ative AI in co-creative processes (Louie et al., 2020).

3 METHODOLOGY

3.1 Context and Outline

A workshop aiming to collaboratively develop a digi-

tal Fairy-Tale discussing positive and negative impli-

cations of AI for the society was conceived. The par-

ticipants were expected to generate music and images

illustrating the plot (text), which was already given.

In Figure 1, children generating music (right) and

images (left) under the supervision of an instructor are

shown. The digital layout of the Fairy-Tale was or-

ganised as shown in Table 1. The children would cre-

ate content for chapters 1 to 4, while the Introduction

and Chapter 5 were already provided as an example.

The workshop was carried out in the context of a

‘Kinder-Uni’, an initiative which tries to engage chil-

dren in scientiﬁc and cultural topics. The workshop

was designed within the ‘Music Pedagogy Project’, a

mandatory course of the Music Education curriculum

whose goal is to design innovative didactic ideas and

test them within a small group of children. Given the

human (2 instructors) and temporal (150 minutes) re-

sources constraints, the workshop was designed for a

maximum of 12 children (aged 10 to 12). Out of the

12 registered ones, 9 participated, eventually.

The workshop was divided into six phases (i-

vi) developed in two different areas of LEONARDO

Zentrum (a research and innovation centre): the Co-

working space (suited for presentations and discus-

sions); and the Laboratory (equipped with the hard-

CSME 2024 - 5th International Special Session on Computer Supported Music Education

624

Table 1: Layout of the Fairy-Tale. The discussed content,

as well as the evoked emotions (in brackets), are indicated.

Section Content

Introduction Describes the characters and context of the plot

Chapter 1 AI to support learning (excitement)

Chapter 2 The risk of privacy violation (anxiety)

Chapter 3 The risk of bias and stereotypes (disappointment)

Chapter 4 AI to understand general trends (anger, calm)

Chapter 5 Importance of ethical regulations

ware and software needed for the hands-on experi-

ment). The six phases can be summarised as follows:

(i) Welcome Message (Co-working space, 5 min):

a brief explanation of formalities including presenta-

tions and the general outline of the workshop.

(ii) Overview of AI (Co-working space, 25 min): ex-

amples of AI applications are described and discussed

through children-friendly materials. Subsequently,

the Introduction of the Fairy-Tale is used as an exam-

ple to illustrate the goals of the workshop, i. e., learn-

ing about AI opportunities and challenges as well as

using AI to generate suitable music and images.

(iii) Co-creative Experimentation with AI (Labo-

ratory, 60 min): the kids, divided in two groups (A

and B) and supervised by each instructor respectively,

work on different chapters of the Fairy-Tale in par-

allel: Group A (chapters 1 − 2); Group B (chapters

3 − 4). After discussing the meaning and emotional

content of a given chapter with the instructor, the chil-

dren work on the generation of songs and images for

that chapter for 30 minutes. To enable satisfactory in-

teractions with the tools, each group is subsequently

split in two sub-groups in a way that a maximum of

3 children interact together at one PC. For instance:

A.1 focusing on the images; A.2 focusing on the mu-

sic. This process is repeated for the other assigned

chapter (which evokes a contrasting emotion, cf. Ta-

ble 1). This time, the children who previously exper-

imented with music would work on the images and

vice versa. The children generate as many images

and songs as they want in the available time; Then,

the group jointly decide on the best option.

Break (10 min): one of the instructors inserted the

musical tracks generated by the children into the dig-

ital Layout (i. e., Canva) and downloaded the project

in video format. Note that the images where directly

generated/inserted by the children in Canva.

(iv) Exchange and Final Discussion (Co-Working

space, 30 min): after projecting each chapter of the

Fairy-Tale, the children engaged in a short discussion

concerning content and usability experience.

(v) Evaluation (Co-Working space, 10 min): the chil-

dren are invited to individually reﬂect on their expe-

rience and to ﬁll out a questionnaire used to assessed

the acquisition of the learning goals (cf. Section 3.2).

(vi) Conclusion (Co-Working space, 10 min): ﬁnally,

after the core part of the workshop has concluded, the

parents are invited to come into the Co-Working space

and the resulting video is watched together.

3.2 Lifelong Learning Competences

The workshop is framed around three lifelong learn-

ing competences (LLC) deﬁned by the European

Commission: Digital and technology-based compe-

tences (LLC4); Active citizenship (LLC6); and Cul-

tural awareness and expression (LLC8). The core ob-

jectives within each LLC are formulated in line with

the UNESCO guidelines (Miao et al., 2023). Con-

cerning LLC4, the workshop aims to promote the

ability to interact with (procedural skills) but also to

understand (conceptual skills) the basics of AI tech-

nologies. Concerning LLC6, the workshop aims to

instil the importance of AI ethics and the need for

humans to become aware of the opportunities and

risks associated with AI. Finally, concerning LLC8,

the workshop aims to promote creative reﬂection, a

form of metacognition which emerges from interac-

tions in novel situations and is shown by the ability to

imagine and describe new creative solutions (Cook,

1998, p. 46). Creative reﬂections would be shown by

children’s ability to imagine how the emotions of a

given chapter could be evoked through musical means

(a process facilitated by the interaction with AI).

For each competence, several learning outcomes,

articulated in knowledge, skills, and attitudes, were

formulated as follows. Concerning LLC4, children

can deﬁne AI, successfully interact with the gener-

ative tools, and demonstrate interest in technology.

Concerning LLC6, children can describe risks asso-

ciated with AI, critically evaluate social unfairness,

and demonstrate cooperative and respectful attitudes

towards others. Concerning LLC8, children can de-

scribe connections between musical parameters and

emotional meanings, can appropriately adapt the mu-

sical output to ﬁt the narrative of the chapter, and are

open to experience new co-creative interactions.

The knowledge, skills, and attitudes to be acquired

can be broadly articulated in four general Learning

Goals: LG1, to gain a ﬁrst understanding and experi-

ence with AI; LG2, to reﬂect about the ethical impli-

cations of AI; LG3, to experiment in evoking different

emotions through music; and LG4, to collaboratively

explore generative AI to expand the own creativity.

3.3 Teaching Methods

Previous research highlights the importance of pro-

moting students’ content creation through collabora-

tive work on multiple technologies to develop digi-

Techville’s Chronicles: A Music Pedagogy Project to Foster Children’s AI Literacy Through Co-Creativity and Multimedia Storytelling

625

tal competences (Lakkala et al., 2011). Thus, col-

laboratively creating digital stories, which has shown

to brings great beneﬁts in the classroom (Robin,

2008), seems to be suitable to develop 21st Century

Skills, too. To reach the described learning goals, a

student-centred teaching method implemented within

a project-based learning setting, was employed. The

digital storytelling approach Stories that inform or in-

struct (Robin, 2008), i. e., digital stories partially de-

veloped by the instructors and presented to the chil-

dren in order to convey speciﬁc information and pro-

mote the acquisition of concrete skills, was used.

In particular, the acquisition of conceptual knowl-

edge, e. g., deﬁning AI, describing its ethical impli-

cations, and reﬂecting on relationships between emo-

tions and musical parameters, as well as critical think-

ing skills were promoted through guided discussions,

also known as modern Socratic method (Le, 2019).

Structured dialogues were lead through the Turn-

Taking-Reading strategy, used to collaboratively read

the story plot whose scenes where alternated with

questions posed by the instructor.

Subsequently, each sub-group was assigned a task,

generating either the missing music or the images.

The hands-on skills and attitudes were promoted

through active learning, i. e., the children were ex-

pected to learn by doing (exploring the tools by them-

selves) while the instructor assumed a facilitator role.

3.4 Generative Tools and Resources

A variety of resources (hardware and software) were

used. Hardware: 1 projector and 1 laptop (used by

one instructor); 4 desktops computers (one per sub-

group, i. e., A.1, A.2, B.1, B.2, cf. Subsection 3.1).

Software: ChatGPT, RunwayML, and Tha3 (used

only by the instructors during the preparation of the

Fairy-Tale layout); Canva and AIVA (used also by the

children during the Workshop). All the AI tools were

accessed through a free account. For security rea-

sons, the Leechblock browser-blocker was installed

to make the children only able to interact with Canva

and the desktop version of AIVA.

ChatGPT

is a chatbot developed by OpenAI de-

signed to enable natural human-machine written con-

versations. ChatGPT was used by the instructors dur-

ing the workshop preparation to generate the Fairy-

Tale plot. Through iterative exchanges, the content,

language style, and text length were deﬁned. In the

ﬁnal plot, it is hardly discernible what was proposed

by ChatGPT and what by the instructors—both inﬂu-

enced each other during the co-creative interaction.

No interaction with ChatGPT during the workshop

https://chat.openai.com

Figure 2: PC-screen during the interaction with Canva aim-

ing to generate an image that represents the given text

was considered as the tool is not expected to be used

by children and due to the informative purpose of the

story, its content had to be designed beforehand.

RunwayML

is a platform to create and trans-

form visual art with the support of generative AI. It

includes text-to-image technology, i. e., models able

to generate images from a textual description. Run-

wayML was used by the instructors during the work-

shop preparation to modify the background of some

images containing the main characters (generated

with Canva, see below). Although the background re-

placement functionality was also experimentally im-

plemented in Canva, its performance was considered

still sub-optimal to be used in the workshop.

Manual Poser Tool from Tha3 - Talking Head

Anime 3 (Khungurn, 2022) is an application that en-

ables to modify anime facial expressions from a given

image in order to portrait different emotions. It was

used by the instructors to modify the facial expres-

sions of the characters over the story. This was needed

to keep them recognisable across the Fairy-Tale since,

due to the non deterministic output of the text-to-

image generation model in Canva, every time a new

text prompt was written, the protagonists would be

differently generated. Given the difﬁculty and time

needed to use this tool effectively, the children were

not expected to interact with it but only with Canva.

Canva

is an online graphic design tool that en-

ables to intuitively create audiovisual content and was

used for the Fairy-Tale layout including text, images,

and music (cf. Figure 2). Canva also integrates a text-

to-image generator, which was used by the children to

create the images. The example images, including the

source for the main characters (before creating further

variations with RunwayML and Tha3) were also cre-

ated by the instructors in Canva with the Anime style.

https://runwayml.com

In English, translated as: AI makes judgements based

on a huge pool of data. The extent to which a data pool is

free of bias determines how fair the AI’s answers are.

https://www.canva.com

CSME 2024 - 5th International Special Session on Computer Supported Music Education

626

Figure 3: PC-screenshot during the interaction with AIVA

aiming to generate a track evoking the chapter’s emotion.

AIVA

is an AI music generation assistant that en-

ables the user to control a variety of elements such

as genre, instrumentation, or chord progression. The

step-by-step option, which leads the user through sev-

eral decision-making steps, was used in the desk-

top version of the tool (cf. Figure 3). At each step,

the children were encouraged to try out different op-

tions, reﬂect on the chapter’s emotional content, and

then make the musical decisions accordingly. The co-

creative interaction with AIVA implied the following

steps: (i) choosing a musical genre to be taken as ref-

erence; (ii) automatically create a chord progression

from a text prompt; (iii) changing the instruments and

altering speciﬁc layers (by removing, re-generating

or manually modifying them) as considered adequate;

(iv) automatically generate and save the ﬁnal track.

For evaluation, an observation protocol to be ﬁlled

out by the instructors and a questionnaire for the chil-

dren were designed. The observation protocol is made

up of eight questions: Q1-Q3 related to the LGs; Q4-

Q8 assessing the appropriateness of the set-up (i. e.,

teaching methods, content, and resources). The chil-

dren questionnaire is divided into ﬁve parts: Sec-

tion 1 (assessing the appropriateness of the set-up);

Section 2 (assessing LG1 and LG2); Section 3 (as-

sessing LG3); Section 4 (assessing LG4); and Sec-

tion 5 (an empty box inviting the children to freely

share their thoughts, something typical of elementary

school inquiries and which the children usually use

to creatively express themselves). The question Do

you want to experiment further with AI?, although in-

terpreted according to LG1, is intentionally included

at the end of Section 4. The reason of this is to en-

courage the children in providing more informed an-

swers which would derive from their reﬂection on the

questions stated over the previous Sections. Common

practices, such as formulating the questions in unam-

biguous and straightforward language, were followed

(Bell, 2007). Due to the preliminary nature of the pi-

lot study as well as constrained resources in terms of

https://www.aiva.ai

duration and number of participants, comparative de-

signs assessing pre- vs post-intervention were explic-

itly avoided in favour of qualitative evaluation and a

descriptive interpretation of rating scales. An English

translation of both instruments is freely accessible.

4 RESULTS

4.1 Instructors’ Perspective

Concerning LG1, the ﬁrst impression was that some

children were already particularly knowledgeable and

interested in AI. The ones more active in the introduc-

tory discussion were able to name AI tools they were

familiar with and were eager to know which tools they

were going to use. Their initial idea about AI was

something related to programming and some kind of

human-generated autonomous entity.

Concerning LG2, children were in principle aware

of some potential risks related to AI, but could not

clearly identify them at the beginning. After the prac-

tical experience, in the discussion (phase iv), they

could at least repeat the speciﬁc risks illustrated in

the Fairy-Tale. Some children could generalise such

risks by giving examples from their own experiences,

e. g., the fact that generally male like dinosaurs does

not mean that all male must like dinosaurs; thus, an

AI that suggests the latter, would be unfair.

Concerning LG3, the importance of the music for

the audiovisual Fairy-Tale was clear for the children,

who could describe in simple terms how music would

convey speciﬁc emotions. Indeed, one of the chil-

dren requested the instructor to combine two gener-

ated tracks in a way the second one would perfectly

ﬁt the last scene of chapter 4, whose generated image

evoked a more modern/active atmosphere (cf. minute

6:18 in the video). Nevertheless, beyond reﬂecting on

general atmosphere/mood, the relationship between

emotional concepts and musical ones could not be ad-

dressed in detail due to the limited amount of time.

Concerning LG4, the children were generally very

interested in the topic and keen to use the computer.

Some where easily satisﬁed by the AI-generated out-

put, while others where much more critical and tried

to modify their actions in order to improve the re-

sults. In the latter case, co-creative interactions were

initiated. In some cases, children very engaged and

perfectionists on one task were easily satisﬁed on the

other, which highlights the importance of considering

children’s interest (beyond tools’ appropriateness) to

enable successful co-creative processes.

In terms of methodological choices, the content

showed to be suitable for the target group. Despite an

Techville’s Chronicles: A Music Pedagogy Project to Foster Children’s AI Literacy Through Co-Creativity and Multimedia Storytelling

627

initial disinterest concerning Fairy-Tales in general,

probably due to the children’s age (they felt too old

for Fary-Tales), children’s attitude was very positive

when interacting with the actual story, whose topic

was timely enough not being perceived as child-like

content. Indeed, the children were actively engaged

during the whole workshop, i. e., not only in the active

learning part (phase iii), but also in the others.

In terms of technological difﬁculty, despite an ini-

tial uncertainty about AIVA, both tools showed to be

appropriate for the target group. The children were

able to naturally interact with them and explore be-

yond what was expected. Some of the children au-

tonomously interacted with AIVA’s midi track until

they obtained the expected outcome. However, the

time necessary to load new instruments and the low

volume of the chord progression pre-view, led to some

frustration. Concerning the images, some children

showed slight disappointment as they seemed to know

more sophisticated software. Still, the fact that they

were unable to generate the main characters of the

Fairy-Tale due to the non-deterministic nature of the

image generator brought up discussions about the dif-

ference between rule-based and stochastic processes.

The children explored this by running several times

the text-to-image generator on a given prompt.

Another consideration to be made is that while

Canva was available in the native language of the chil-

dren, i. e., German, AIVA was only available in En-

glish. This was hypothesised to limit children’s inter-

action, which was the case for some of them, although

not for all: thus, highlighting the need of tools in lan-

guages beyond English.

Finally, the planned time was slightly short, as

some children intentionally decided not to make the

break in order to reﬁne the generated music/images.

In addition, since the ﬁnal discussion (phase iv) was

longer than expected, the time dedicated to the eval-

uation (phase v) had to be reduced. Due to this,

some children were not able to ﬁnish the question-

naire and requested to do it after watching the video

with the parents (phase vi). In terms of group dynam-

ics, the sub-group interaction was optimal with two

children as this allowed an easier communication and

decision-making while working in groups of three left

one child to some extent apart.

4.2 Children’s Perspective

Concerning LG1, the answers to the questionnaire

show that most of the children were able to provide

a plausible deﬁnition of AI, indicating that it is a

technology created by humans which has the ability

to learn. Some of them did not provide an actual

Figure 4: Multiple-choice responses related to LG3.

deﬁnition but rather said for what it might be used,

which suggests that addressing the topic more for-

mally would be needed. All of them indicated they

have heard about AI before the workshop, have under-

stood the tasks to be performed, and highlighted their

motivation to work again with AI in the future. To the

question What did you ﬁnd bad in the workshop? all

answered ‘nothing’ except for one who indicated that

there could have been more variety in the activities.

This however, contrasts with the general opinion that

the time was too short (6 children would have liked

more time, 3 considered that the time was sufﬁcient).

Concerning LG2, all children were able to provide

precise examples of the chances and risks. Concern-

ing examples of how AI can support humans, they

generally (6 of them) considered as most important

that AI can support learning and acquiring knowl-

edge, while one of them mentioned as beneﬁt some-

thing not discussed in the workshop, i. e., the pos-

sibility of using AI to perform tasks dangerous for

humans, such as space travel. Finally, one child in-

dicated that AI would not imply any risk if we take

appropriate precautions while the others considered

that using AI might be associated to some risks. They

were able to repeat those discussed within the Fairy-

Tale, and even formulate new ones, e. g., military use.

Concerning LG3, all the children agreed on the

importance of music for the Fairy-Tale, but only 5 in-

dicated explicitly that it was needed to create the ap-

propriate atmosphere/emotional situation. The music

was perceived as easier to generate w. r. t. the images

(cf. 7 vs 2 in Figure 4, above). This is probably due

to the more abstract options the children imagined for

the music compared to the images. Indeed, concern-

ing the images, the children explicitly referred to the

difﬁculty of exactly generating what they had in mind.

Thus, in terms of children’s level of satisfaction with

the AI output, a relationship between their interest and

perceived simplicity could be drawn: generating the

music was clearly perceived as simpler and was also

slightly seen as more interesting (cf. Figure 4, below).

Indeed, two children (one of them mostly enjoying the

image task) explicitly expressed that the most inter-

CSME 2024 - 5th International Special Session on Computer Supported Music Education

628

Figure 5: Self-perceived conﬁdence improvement (related

to LG4). X-axis: Children’s conﬁdence with tools similar

to those used in the workshop; Y-axis: Number of answers.

esting task was the one implying less effort (simpler).

Finally, since successful co-creative interactions

might depend on the self-perceived agency, to evalu-

ate LG4, the children were requested to indicate their

general level of conﬁdence with tools like those used

in the workshop (very conﬁdent, conﬁdent, not so con-

ﬁdent)

as well as their self-perceived improvement

due to the workshop (no, to some extent, yes). The re-

lationship between these two questions is shown in

Figure 5. The children feeling very conﬁdent with

computers (5 out of 9) showed a lower self-perceived

improvement: 2 of them indicated no, i. e., they did

not feel more conﬁdent after the workshop, 3 indi-

cated the workshop helped to some extent. On the

contrary, those feeling not so conﬁdent showed a pos-

itive trend: one indicated a clear improvement (yes),

the other to some extent. The remaining two children,

both considering themselves conﬁdent, perceived a

clear improvement in their own skills after the work-

shop. Although no relationship between perceived

improvement and own creativity development can be

drawn, perceived learning can be seen as a sign of

positive attitude towards the co-creative experience.

Finally, 3 of the children indicated as particularly

good the possibility to do work themselves. This was

the case for 2 children identifying themselves as very

conﬁdent as well as for 1 not so conﬁdent, and sug-

gest a relationship between active learning and chil-

dren’s enjoyment/motivation. However, the child who

self-identiﬁed as not so conﬁdent attributed the lim-

ited self-perceived improvement to the fact that dur-

ing the workshop there was not so much explained.

This shows that even if, as expected, active learning

is appreciated by the children, providing sufﬁcient ex-

planations is needed to guarantee meaningful learn-

ing that goes beyond simple interaction, especially for

children with a weaker technical background.

Note that an additional option indicating absolutely not

conﬁdent was given but none of the children chose it.

5 DISCUSSION

Children’s positive reaction to the music task contra-

dicts to some extent previous works showing users’

overwhelming feelings as a reaction to their limited

involvement w. r. t. the amount of generated musical

content (Louie et al., 2020). A reason for this differ-

ence might be the age of the participants: although

Louie et al. (2020) do not indicate it, by the descrip-

tion it seems they are at least young adults. Thus,

older participants, who had the time to acquire a mu-

sical background, would naturally have more expec-

tations and therefore would be disappointed by an AI

generating too much content which does not mirror

their speciﬁc expectations. This was indeed the case

for the children concerning the images, who were dis-

appointed by the non-deterministic output, something

also pointed out by (Louie et al., 2020).

Previous works have shown that children would

change the own attitude towards AI when discovering

that it can speak their native language (Druga et al.,

2019). The workshop’s outcomes show that while the

language did not play a role in the co-creative pro-

cess for some children, interacting in a foreign lan-

guage limited the experience of others. Even for those

whose interaction quality seemed to be independent

of the language, this is only applicable to the investi-

gated scenario. Just because some children’s English

skills were sufﬁcient to explore the interface does not

mean that they would have shown the same attitude

if the interaction had been carried out orally, as in

(Druga et al., 2019). Furthermore, during the inter-

action with AIVA, the children were only expected

to understand the name of some dialogues in English

while they could still insert text prompts in German:

this was explicitly requested by the children.

6 CONCLUSIONS

The presented pilot study suggests that the integration

of AI-child co-creativity within a digital storytelling

framework might be a promising method to promote

several lifelong learning competences. In particular,

music generative tools such as AIVA, with functional-

ities that encourage users’ interaction, are an interest-

ing option to support children’s creativity while im-

plicitly promoting AI-literacy. In addition, the pre-

sented framework might offer educational advantages

with respect to traditional storytelling, since besides

its inherent ﬂexibility towards any instructional con-

tent, it also integrates a powerful active learning com-

ponent, i. e., AI-supported co-creativity.

Despite the positive feedback from the children,

Techville’s Chronicles: A Music Pedagogy Project to Foster Children’s AI Literacy Through Co-Creativity and Multimedia Storytelling

629

limited discussion about the music-emotion connec-

tions was observed. This was in part due to time con-

straints but also due to the fact that the topic was not

formally addressed in the introduction (which was fo-

cused on AI). Due to the complexity of the topic (to

some extent underestimated), exploring musical emo-

tions through co-creative interactions with generative

AI should be addressed on a speciﬁc workshop.

Finally, it goes without saying that due to the small

evaluated sample, these outcomes cannot be gener-

alised beyond the investigated group. Similarly, ded-

icating several sessions to the topic would be natu-

rally beneﬁcial. The presented project should be taken

rather as a proof of concept illustrating how AI-child

co-creativity can be used to promote AI-literacy, even

in a compact format. This might eventually trigger

replication studies in other contexts as well as pro-

mote the engagement of schools in the longitudinal

implementation of such projects, something needed

for an empirical evaluation. Despite its limitations,

the qualitative and methodological discussions result-

ing from this experiment aim above all to rise the at-

tention on a important (under-researched) topic and

by this, eventually inspiring future related works.

ACKNOWLEDGEMENTS

I am immensely thankful to the children for their en-

thusiasm, which is my biggest motivation to explore

new horizons in teaching and learning. Special thanks

go to my student Judith and all others involved.

REFERENCES

Bell, A. (2007). Designing and testing questionnaires for

children. Journal of Research in Nursing, 12:461–469.

Bell, J. and Bell, T. (2018). Integrating computational think-

ing with a music education context. Informatics in

Education, 17(2):151–166.

Cook, J. (1998). Mentoring, metacognition and music:

Interaction analyses and implications for intelligent

learning environments. International Journal of Ar-

tiﬁcial Intelligence in Education, 9:45–87.

Druga, S., Vu, S. T., Likhith, E., and Qiu, T. (2019). Inclu-

sive AI literacy for kids around the world. In Proc. of

FabLearn, pages 104–111, New York, NY, USA.

European Commission (2019). Key Competences for

Lifelong Learning. European Commission and

Directorate-General for Education, Youth, Sport and

Culture (Publications Ofﬁce: doi/10.2766/569540).

Ferrari, A. and Punie, Y. (2013). DigComp: A framework

for developing and understanding digital competence

in Europe. JRC Scientiﬁc and Policy Report, Publica-

tions Ofﬁce of the European Union, Luxembourg.

Jesionkowska, J., Wild, F., and Deval, Y. (2020). Active

learning augmented reality for STEAM education: A

case study. Education Sciences, 10(8):1–15.

Kantosalo, A., Ravikumar, P. T., Grace, K., and Takala, T.

(2020). Modalities, styles and strategies: An interac-

tion framework for human-computer co-creativity. In

Proc. of ICCC, pages 57–64, Online.

Khungurn, P. (2022). Talking head( ? ) anime from a single

image 3: Now the body too. http://pkhungurn.github

.io/talking-head-anime-3/. Accessed: 16-01-2024.

Lakkala, M., Ilom

aki, L., and Kantosalo, A. (2011). Which

pedagogical practices and methods best support learn-

ing digital competences? Linked portal.

Le, N.-T. (2019). How do technology-enhanced learning

tools support critical thinking? Frontiers in Educa-

tion, 4:1–9.

Liao, C. (2016). From interdisciplinary to transdisciplinary:

An arts-integrated approach to STEAM education. Art

Education, 69(6):44–49.

Louie, R., Coenen, A., Huang, C. Z., Terry, M., and Cai,

C. J. (2020). Novice-AI music co-creation via AI-

steering tools for deep generative models. In Proc. of

CHI Conf. on Human Factors in Computing Systems,

pages 1–13, Honolulu, HI, USA.

Miao, F., Giannini, S., and Holmes, W. (2023). Guidance

for Generative AI in Education and Research. UN-

ESCO, Paris, France.

Ng, D. T. K., Luo, W., Chan, H. M. Y., and Chu, S. K. W.

(2022). Using digital story writing as a pedagogy to

develop AI literacy among primary students. Comput-

ers and Education: Artiﬁcial Intelligence, 3:100054.

Ozer, Z. and Demirbatir, R. E. (2023). Examination of

STEAM-based digital learning applications in music

education. European Journal of STEM Education,

8(1):1–11.

Robin, B. R. (2008). Digital storytelling: A powerful tech-

nology tool for the 21st century classroom. Theory

into Practice, 47(3):220–228.

Striano, F. (2019). Towards “post-digital”. A media theory

to re-think the digital revolution. Ethics in Progress,

10(1):83–93.

Su, J. and Yang, W. (2022). Artiﬁcial intelligence in early

childhood education: A scoping review. Computers

and Education: Artiﬁcial Intelligence, 3:100049.

Tatar, K. and Pasquier, P. (2019). Musical agents: A typol-

ogy and state of the art towards musical metacreation.

Journal of New Music Research, 48(1):56–105.

Vuorikari, R., Kluzer, S., Punie, Y., et al. (2022). Dig-

Comp 2.2: The Digital Competence Framework for

Citizens - With new examples of knowledge, skills and

attitudes. Publications Ofﬁce of the European Union,

Luxembourg.

Williams, R., Park, H. W., Oh, L., and Breazeal, C. (2019).

Popbots: Designing an artiﬁcial intelligence curricu-

lum for early childhood education. In Proc. of AAAI,

pages 9729–9736, Honolulu, HI, USA.

Zhang, K. and Aslan, A. B. (2021). AI technologies for edu-

cation: Recent research & future directions. Comput-

ers and Education: Artiﬁcial Intelligence, 2:100025.

CSME 2024 - 5th International Special Session on Computer Supported Music Education

630