Nikola Novaković, Damjana Kolednjak – Does It Sparkle? The AI-Generated Book Alice and Sparkle (2022) in the Context of Picturebook Research

Introduction

In a Twitter ^{The platform has since been rebranded as “X”.} post from 9 December 2022, Amaar Reshi, a product designer based in Silicon Valley, described how he “spent the weekend playing with ChatGPT, MidJourney, and other AI tools… and by combining all of them, published a children’s book co-written and illustrated by AI!” (Reshi 2022a). In the photograph attached to the post, Reshi is seen smiling and holding up an early version of the book, entitled Alice and Sparkle. The back cover of the 2023 hardback edition of the book contains the following summary:

This is a story about a young girl named Alice who discovers the magic of artificial intelligence. She creates her own AI, named Sparkle, and together they go on adventures and use their combined knowledge to make the world a better place. The story explores the incredible abilities of AI and the importance of using them for good. It is a tale of friendship and exploration, filled with magic and wonder (Reshi 2023).

Initially, the responses of other netizens were favourable, calling Reshi’s effort “amazing”, “innovative”, and “genius”. Commenters drawn to the post were interested in the particulars of the process and inquired about the exact tools Reshi employed as well as the prompts he fed ChatGPT in order to obtain the story and character descriptions he would subsequently use in MidJourney, the AI image generator, to produce the images for the book. However, it did not take long for those critical of both Reshi’s statements about authorship ^{Among them the following, posted on the same day, 9 December 2022, drew ample ire: “And BOOM. In a weekend, from idea, to illustrations, to becoming a publisher [sic] author!” (Reshi 2022b).} and the ethical implications of his results to begin voicing their opinions. While some attacked Reshi for stealing from artists by disregarding the issue of copyrights and consequently encouraging his audience to imitate his methods, others raised the problem of licensing the results produced by AI generators: would copying Reshi’s book and printing it under another person’s name infringe on any possible copyright Reshi might have over an AI-generated product? The outrage over Reshi’s work spilled over onto the Amazon product webpage for Alice and Sparkle, where, among numerous positive ratings, an equally sizeable portion of disgruntled readers left some strongly worded reviews, calling the book “abysmal”, “creepy”, and “devoid of any spark of creativity,” therefore joining the Twitter commenters in criticizing both the lacklustre storytelling and the flaws in the AI-generated images. Although Reshi claims he was motivated to create the book as “a demonstration of the technology and a gift to the first-borns of my two of my best friends [sic]” (Reshi Ammaar 2024), and originally offered to gift copies of the book to members of his audience on Twitter if they had children, ^{In a separate Twitter post from 9 December 2022, Reshi stated: “And if you have any kids you want to gift this to, let me know! I’d love to send you a copy!” (Reshi 2022b). Reshi’s wording is somewhat ambiguous, especially because the message is included in a post advertising the Amazon product page for Alice and Sparkle, which is accompanied by an image showing the price of the book.} his audience’s vitriolic responses (which Reshi says included death threats (Stokel-Walker 2022)) apparently prompted him to soon attempt to address the outcry: on 12 December 2022, Reshi posted on Twitter that there “are serious, incredibly valid concerns from artists and writers about all this technology” (Reshi 2022c). Striving to redirect the conversation about the source of the problem back to the AI tools he employed, Reshi then posed a question to the creators of the “tools at OpenAI & MidJourney: how do we ensure protections for artists / train models on consent? Their talent, skill, hard work to get there needs to be respected.” (Reshi 2022d). Finally, in a post concluding his statements on the issue, Reshi wrote that “[t]his was all a test of what’s possible,” but that he would “be waiting to hear more on the above before proceeding to create more content” (Reshi 2022e).

Regardless of Reshi’s subsequent efforts at mollifying concerns over artistic theft, the controversial Alice and Sparkle has entered the landscape of children’s literature and inevitably affected its current status, not least by still being sold on Amazon, despite the efforts of Reshi’s Twitter detractors to convince him to remove the book from the online store. Alice and Sparkle thus joins other books intended for children which have been produced using AI assistance, such as the picturebook Robot Ranger: Robot Ethics for Children (2020), written by William Barry, illustrated by Cressy Tylavsky, and co-written by the AI android Maria Bot; If Animals Had Jobs (2022), written by Dennis DeRobertis and illustrated with images generated by DALL-E 2 (and, interestingly, published less than a month before Alice and Sparkle caused such an uproar); Bedtime Stories: Short and Sweet, for a Good Night’s Sleep by Kamil Banc (2022), written by ChatGPT and illustrated by an AI image generator; and Anitha Rathod’s Tick-Tock Tantrums (2023) in which, according to its Amazon page, both the images and the “rhyming style of the story” are created by AI tools. Furthermore, a range of specialised websites now offer the possibility to create one’s own children’s book with the help of artificial intelligence and to have it printed and delivered for a price (see BookBildr 2024; Child Book 2024; Once Upon a Bot 2024), while other sites provide instructions on how to use freely available software for the same purpose (see e.g., Book 2023; Stroh-Spijer 2023). However, while authors like Reshi, Barry, Rathod, and DeRobertis advertise their books as intended for children, and while online book creation services promise to turn the “time-consuming and daunting task” of writing children’s books into something one can accomplish in “10 minutes” (Book 2023), we believe it becomes necessary to examine whether such books actually function as children’s books. Specifically, we ask whether Alice and Sparkle, advertised as a picturebook and here taken as a case study of a text co-written and illustrated by AI under the direction of human intelligence, fits into the genre of picturebooks. To answer this question, we examine it against some of the widely accepted theories of picturebooks (e.g., Bader 1976; Nikolajeva, Scott 2006; Nodelman 1988) and investigate its portrayal of character, narrative perspective, the (non-existent) multimodality of its verbal and visual discourses, and the elements of its peritext and epitext. Most importantly, we focus on the ways in which Alice and Sparkle performs the interdependence of the verbal and visual discourses as the multimodal relationship central to the mechanics of “a picturebook proper”: the interaction of the two expressive means in the process of meaning-making, whose complex dynamics enable the overall impact of the work (Nikolajeva 2006: 107). In other words, does it sparkle?

Confusion of Form

Before Alice and Sparkle can be studied as an example of an interface between artificial intelligence and children’s literature, and before a methodological framework can be selected for that purpose, it is necessary to first define the book’s formal and genre characteristics. However, this is where the first major problem emerges: what, in fact, is Reshi’s book? How does it present itself, how is it advertised, and, no less importantly, how is it perceived by those purchasing, reading, and critiquing the book? While a simple (and, indeed, overly simplistic) answer might be that it presents itself as a book with prominent images and little text, and that it should therefore be classified as a picturebook, such a definition would not only violate some of the basic tenets of picturebook theory, but also obfuscate the ways readers at large perceive the form and genre of Reshi’s book. If we examine the book’s peritext, the description on its back cover mostly focuses on the elements of story, outlining the main protagonist and her “adventures” while merely calling Alice and Sparkle a “book”. Turning to its Amazon store page, we find that the description included on the website attempts only a slightly narrower definition by calling Alice and Sparkle a “children’s book” and “a children’s story at heart.” Going further beyond the confines of the book itself and its advertisement, we turn to news articles that have covered the phenomenon of Reshi’s book. Interestingly, some of these include attempts at a formal definition by either calling Alice and Sparkle “an illustrated book” (Nolan 2023) or a “picture book” (Popli 2022; Brown 2023). This confusion between the two forms may not be accidental. In fact, it is quite common in general, “as the picturebook and the illustrated book are often confused for one another” (Bird, Yokota 2017: 281). The two forms, however, are distinguishable from each other by the nature of the relationship between their visual and verbal discourses. According to Bettina Kümmerling-Meibauer, the term “an illustrated children’s book” “applies to any children’s book containing at least one illustration in any artistic medium” (Kümmerling-Meibauer 2006: 276), but the particular relation of the pictures to the verbal text plays an important role, since “[t]he pictures in an illustrated book must have relevance to the text but should one remove the illustrations, the writing would stand alone” (Bird, Yokota 2017: 281). The illustrations in the picturebook, on the other hand, are indivisible from the verbal text and therefore “serve a primary meaning-creating function” (Bird, Yokota 2017: 282). In a narratological analysis of the picturebook form, Smiljana Narančić Kovač shows how the picturebook “maintains a distinction of the verbal and the visual discourse and of their respective narrators, who mediate the same story” (Narančić Kovač 2017: 411).

Given the differences between the illustrated book and the picturebook, which of the two does Alice and Sparkle belong to? The question is difficult to answer, since Reshi’s book displays characteristics of both forms, occasionally simultaneously. If picturebooks are “characterized by an equal relationship between pictures and words” (Kümmerling-Meibauer 2006: 276), and if pictures cannot be easily removed without impacting both the narration of the story and, consequently, the interpretation of the text, Alice and Sparkle is more easily classified as a children’s illustrated book. In fact, the pictures do not in any clear way contribute to the meaning of the verbal text and could, conceivably, stand alone. In other words, removing them from the book would not impact the understanding of the story conveyed by the verbal discourse. Another feature commonly found in illustrated books are titles or complete sentences, sometimes taken from the verbal text itself, and located under the illustrations, which are placed on a separate page facing a page of verbal text. This convention of repeating the same (or slightly modified) verbal text underlines the fact that the illustrations in an illustrated book “may enhance, decorate, and amplify the text, but the narrative is not reliant on their presence” (Bird, Yokota 2017: 281). This convention is only partially replicated in Alice and Sparkle, as the titles under the images (when they appear) are not rephrasing or repeating sections taken from the verbal text but instead introduce entirely new information. One example of this is the title on page 15 (“And Sparkle could change into all different kinds of robot shapes too!”), which also conveniently explains away the issue of inconsistency in the visual depictions of Sparkle, a result of limitations of AI image generation that Reshi faced while selecting the images for the book.

While the above seems to indicate that Alice and Sparkle can be more easily categorized as an illustrated book, the lack of consistency in including titles under the images, as well as some of Reshi’s other design decisions, suggest that the book should in fact be read as a picturebook. For instance, Perry Nodelman’s definition of picturebooks as “books intended for young children which communicate information or tell stories through a series of many pictures combined with relatively slight texts or no texts at all” (Nodelman 1988: vii) seems easily applicable to Alice and Sparkle, as it both addresses a young audience and achieves Nodelman’s stated balance between text and image. However, Barbara Bader’s definition of picturebooks shows how problematic it is to reliably classify Reshi’s book. Bader states that “[a]s an art form [the picturebook] hinges on the interdependence of pictures and words, on the simultaneous display of two facing pages, and on the drama of the turning of the page” (Bader 1976: 1). Addressing the “interdependence of pictures and words” in their book How Picturebooks Work, Nikolajeva and Scott explain how the reading of picturebooks necessarily involves turning back and forth from the verbal to the visual text. As they say, “each new rereading of either words or pictures creates better prerequisites for an adequate interpretation of the whole” (Nikolajeva, Scott 2006: 2). Nikolajeva and Scott point out that children seem to know this because they demand that the same picturebook be read aloud to them over and over again as “they go more and more deeply into its meaning” (Nikolajeva, Scott 2006: 2). Importantly, they also point out that “too often adults have lost the ability to read picture books in this way, because they ignore the whole and regard the illustrations as merely decorative” (Nikolajeva, Scott 2006: 2). This is what seems to be at work in Reshi’s book, as the AI-generated images do not seem to add anything meaningful to the text, instead performing a simply ornamental role.

Bader’s second principle, that of setting up an expectation or a tension which is then resolved by turning the page, is either largely absent from Reshi’s book or, when it is established, ultimately leads to a frustrating digression that leaves the tension unresolved, but not in a suggestive, meaning-making way. For instance, the arguably most dramatic moment in the story, when the reader is told that “Alice was both proud and scared of her creation” (Reshi 2023: 18), hints at a darker side to Reshi’s story, one which is earlier foreshadowed when we are informed that Alice was aware that AIs “were powerful and could be used for good or evil, depending on how they were guided” (Reshi 2023: 9). However, the drama of Alice’s conflicting emotions of pride and fear is abandoned by the turning of the page, and replaced by an abrupt and entirely vague ending: “Together, Alice and Sparkle went on many adventures and used their combined knowledge to make the world a better place” (Reshi 2023: 20). Moreover, the two images on pages 19 and 21 look almost identical in composition, thus in no way supporting the drama established in the accompanying verbal text.

The images in Reshi’s book also fail to indicate a relationship with the verbal text which we could describe as symmetrical, complementary, or contradictory – the terms that Nikolajeva and Scott use to describe some of the possible relationships between verbal and visual texts. Whereas in some of the picturebooks explored by Nikolajeva and Scott “[b]oth words and images leave room for the readers/viewers to fill with their previous knowledge, experience, and expectations” (Nikolajeva, Scott 2006: 2), Reshi’s verbal text leaves gaps that images never fill. Another moment in Reshi’s book that seems imbued with potentially dramatic implications occurs when we are told that “as Sparkle grew more powerful, it began to make its own decisions. It started to explore the world by itself and learn from those experiences” (Reshi 2023: 16). Again, however, the AI-generated image in no way completes the gaps in the story, never answering the question of which decisions the robot Sparkle made, or how it explored the world. Instead, the accompanying image of Alice and Sparkle posing for the viewer is almost identical to the two succeeding images, thus underlining a clear demarcation between the story narrated in the verbal discourse and the absence of a narrative in the purely decorative, visual discourse.

Unintentional Metafictionality

And yet, despite the absence of a narrative in the visual component of Alice and Sparkle, we wish to contend that its visual discourse does, in fact, tell a story, albeit not necessarily the one originally intended by the author and his AI assistants. What Alice and Sparkle seems to be communicating and commenting on, far more than the rather superficial story of the girl and her robot, is in fact the process of its AI-assisted production. This first emerges in the epitextual context of Reshi’s Twitter posts, in which he comments on the issue of achieving a consistent style across AI-generated images, particularly in the depiction of the characters. While Alice may, at first glance, seem to be portrayed in a somewhat constant manner (in all of the images she is shown as Caucasian, with curly hair, a slender build, and invariably wearing dresses of a Western style), a second glance reveals several inconsistencies: her eye colour shifts between blue and brown; her age seems to vary; and her facial structure changes from one image to another. To make matters even more complicated, Alice is portrayed in four different variations on page 13, which carries a composition of four images. Reshi himself addresses this problem in the afterword, where he says that “this book also has imperfections: inconsistent art styles and slight issues in the images” (Reshi 2023). The “slight issues in the images” which Reshi refers to are in fact various mistakes in the generation of images resulting from the AI’s attempt to merge together various images based on the provided queries. For instance, there are issues with the generation of Alice’s eyes (p. 8 and 13) and fingers (p. 8, 10, and 17), as well as with her clothing and the objects and people surrounding her (with some being particularly uncanny, as on p. 21, where a child in the background seems to be holding the hand of a headless woman).

In order to test the assumptions about Alice and Sparkle – those considering it a picturebook as well as those seeing it as a book which primarily, yet at least in part unintentionally, comments on its status as a product – a read aloud of Alice and Sparkle was conducted in May 2024 with five 10-year-old primary school pupils (three girls and two boys) from Croatia. The children were not told about the involvement of AI in the production of the book until they independently noticed the signs of AI use. Over the span of approximately one school lesson (44 minutes), the children discussed both the story of Alice and Sparkle and how it is conveyed to the reader. When asked to describe the story and its characters, the children quickly agreed that the book ended too abruptly and that “the author doesn’t say anything,” with one boy adding that “the book doesn’t really have any kind of meaning to it.” Interestingly, when the children were asked to describe the illustrations, one boy immediately said he “realized that these pictures are probably AI-generated,” using the differences in the depictions of Alice to support his claim: “Like, that picture looks completely different from the other picture, she looks younger, happier…” Other children immediately agreed. One of the boys in the group pointed out the differences between the depictions of Alice inside the book and the Alice from the cover illustration, while another boy focused on the robot, saying that it “keeps on changing colours.” Later in the discussion, the mistakes in AI-generation drew further comments from the children, leading them to call Alice “creepy.” In fact, the children’s impressions almost exclusively focused on the details of the book’s production and its failings in conveying a fulfilling narrative, particularly as a result of the disjunction between the verbal text and the repetitive images.

What the read aloud with the group of children confirms is that Alice and Sparkle consistently, but not necessarily intentionally, draws attention to itself as a product of AI generation. As mentioned above (and as restated by the group of children), the absence of a multimodal narrative, which would emerge from a close connection between the verbal and visual discourses, invites a reading of the book as a story of its own AI-assisted creation. As a result, we wish to propose that Alice and Sparkle is an example of a book with a partially unintentional metafictional effect, therefore diverging from the conventional definitions of metafiction. In her chapter of The Routledge Companion to Picturebooks, “Picturebooks and Metafiction,” Maria Cecilia Silva-Díaz, drawing on Patricia Waugh’s seminal work on the subject, defines metafiction as fiction which has the capacity to reflect on its own ontological status as an artefact or a construction, and that it can perform it both in an explicit and implicit way (Silva-Díaz 2017: 73). Reshi’s book already features a level of intentional implicit metafictionality in his presentation of a story about artificial intelligence, an effect which is established by aligning Reshi’s reliance on ChatGPT and Midjourney with Alice’s use of AI. However, a second level of the book’s metafictionality, one that we would term both implicit and unintentional, emerges partially from the mistakes in the generated images, but mainly from the absence of a close relationship between the illustrations and the text as a characteristic expected of a picturebook. As a result, this effect, which relegates to the illustrations’ merely decorative role despite seemingly positioning itself within the boundaries of the picturebook form, draws the attention of the reader not to their involvement in the “reconstruction of the sense” of the book (Silva-Díaz 2017: 72), but to their inability to engage in this reconstruction at all. If, as one of the children commented, the “book doesn’t really have any kind of meaning to it,” the read along showed that this “quality” of the book invites the reader to examine how its absence of meaning is achieved. As the children discovered, its key resides partially in the inconsistency of the illustrations, but mainly in the process of violating the sense-making role of the visual discourse in the expected text-image dialogism of a picturebook.

Conclusion

In a sense, Reshi’s book as a material object becomes the centre of attention, thereby simultaneously replicating the conditions of its revelation on Twitter, which would become mired in controversy due to its AI origins. As the children’s comments showed, Reshi’s book does not invite us to think about its rather underdeveloped fictional world and almost non-existent characters as much as it encourages us to ponder such issues as authorship shared with AI, copyrights, picturebook design, and, finally, the definition of metafiction. In fact, the children’s responses mirror those of other readers who have left their comments on the book’s Amazon web page. Instead of being drawn in by the story and the relationship between the verbal and the visual discourses, which remain strictly divided, never borrowing from each other’s specific strategies, and therefore never becoming multimodal in a manner often found in highly-regarded picturebooks (Narančić Kovač 2015), readers generally seem to focus on the book as a product, the process and result of the book’s creation, and on the intentions of the author. And despite Reshi’s hopes in the afterword to Alice and Sparkle claiming that his book will lead “to a series of empowering tools that are built responsibly, benefiting and enabling a new set of creators while also protecting those that inspired all of them” (Reshi 2023), Alice and Sparkle (and the headlines surrounding its publication) may have played a role in the growing industry of AI-generated books, some of which now seem intent on avoiding the disclosure of AI involvement and thus, perhaps, the backlash that Reshi’s book and he himself have been exposed to: books by such authors as Finja Schön (Emily and the Mermaid Stone (2023)); the prolific J.P (sic) Anthony Williams (Lucy and the Enchanted Forest: Bedtime Story Book for Kids (2023)); and Bold Kids, the “author” and publisher of more than 500 books available on Amazon which are more or less obviously written by AI (with Sheeps: Children’s Book Filled with Facts (2022) being the most egregious example).

What unites all such titles, whether the identity of their AI co-authors is disclosed or not, is a potentially inevitable distraction which pulls the reader away from immersing in the storyworld and toward scrutinising any noticeable traces of AI involvement. The result of this is a situation which challenges us to rethink not only our definition of metafictional effects as necessarily intentional, but also the picturebook form itself. The case of Alice and Sparkle urges those of us working in the field of picturebook research to make more room for an exploration of a specific subset of the picturebook as a new form emerging from AI (co-)authorship, and to seek answers to the question of how various artificial intelligences, as sometimes credited authors and occasionally undisclosed collaborators, increasingly affect the shifting landscape of picturebook production, consumption and evaluation.

References:

Bader, Barbara (1976). American Picturebooks from Noah’s Ark to the Beast Within. New York, London: Macmillan.

Banc, Kamil (2022). Bedtime Stories: Short and Sweet, for a Good Night’s Sleep. Independently published.

Barry, William; Tylavsky, Cressy; Bot, Maria (2020). Robot Ranger: Robot Ethics for Children. Creative Society Media.

Bird, Elizabeth; Yokota, Junko (2017). “Picturebooks and Illustrated Books.” The Routledge Companion to Picturebooks (Kümmerling-Meibauer, Bettina, ed.). New York: Routledge, 2017, pp. 281–290.

Bold Kids (2022). Sheeps: Children’s Book Filled with Facts. Bold Kids.

Book, Adrien (2023). “Write a Children’s Book in 10 Minutes with AI.” The Pourquoi Pas, 2 January, https://www.thepourquoipas.com/post/write-a-childrens-book-quickly-with-ai. Accessed 29 May 2024.

BookBildr (2024). https://www.bookbildr.com/. Accessed 29 May 2024.

Brown, Alesha (2023). “Making Artificial Intelligence Work for You.” Publishers Weekly, 25 August, https://www.publishersweekly.com/pw/by-topic/authors/pw-select/article/93054-making-artificial-intelligence-work-for-you.html. Accessed 29 May 2024.

Child Book (2024). https://www.childbook.ai/. Accessed 29 May 2024.

DeRobertis, Dennis (2022). If Animals Had Jobs. New York: Stone Hollow Press.

Kümmerling-Meibauer, Bettina (2006). “Illustration.” The Oxford Encyclopedia of Children’s Literature, vol. 2. (Zipes, Jack, ed.). Oxford: Oxford University Press, pp. 276–281.

Narančić Kovač, Smiljana (2015). Jedna priča – dva pripovjedača. Zagreb: ArTresor naklada.

Narančić Kovač, Smiljana (2017). “Picturebooks and Narratology.” The Routledge Companion to Picturebooks (Kümmerling-Meibauer, Bettina, ed.). New York: Routledge, pp. 409–419.

Nikolajeva, Maria (2006). “Word and Picture.” Teaching Children’s Fiction (Butler, Catherine, ed.). Basingstoke, New York: Palgrave Macmillan, pp. 106–151.

Nikolajeva, Maria; Scott, Carole (2006). How Picturebooks Work. New York: Routledge.

Nodelman, Perry (1988). Words About Pictures: The Narrative Art of Children’s Picture Book. Athens, GA: University of Georgia Press.

Nolan, Beatrice (2023). “This Man Used AI to Write and Illustrate a Children’s Book in One Weekend. He Wasn’t Prepared for the Backlash.” Insider, 15 January, https://www.businessinsider.com/chatgpt-midjourney-ai-write-illustrate-childrens-book-one-weekend-alice-2023-1. Accessed 29 May 2024.

Once Upon a Bot (2024). https://onceuponabot.com. Accessed 29 May 2024.

Popli, Nik (2022). “He Used AI to Publish a Children’s Book in a Weekend. Artists Are Not Happy about It.” Time, 14 December, https://time.com/6240569/ai-childrens-book-alice-and-sparkle-artists-unhappy/. Accessed 29 May 2024.

Rathod, Anitha (2023). Tick-Tock Tantrums. Independently published.

Reshi Ammaar (2024). https://ammaar.me/. Accessed 29 May 2024.

Reshi, Ammaar (@ammaar) (2022a). “I spent the weekend playing with ChatGPT, MidJourney, and other AI tools… and by combining all of them, published a children’s book co-written and illustrated by AI! Here’s how!” 9 December, 7:35 p.m. Tweet, https://twitter.com/ammaar/status/1601284293363261441. Accessed 29 May 2024.

Reshi, Ammaar (@ammaar) (2022b). “And BOOM. In a weekend, from idea, to illustrations, to becoming a publisher author! You can check out the book on Amazon here. US: https://a.co/aN0IVPx UK: https://amzn.eu/d/0sjE7lC And if you have any kids you want to gift this to, let me know! I’d love to send you a copy!” 9 December, 7:35 p.m. Tweet, https://twitter.com/ammaar/status/1601284455410585600. Accessed 29 May 2024.

Reshi, Ammaar (@ammaar) (2022c). “Adding this to the main thread after a lot of helpful convos with folks willing to discuss—this is a PSA! There are serious, incredibly valid concerns from artists and writers about all this technology. Their emotional responses are fair, we should listen and instead should ask:” 12 December, 5:20 p.m. Tweet, https://twitter.com/ammaar/status/1602337682138750978. Accessed 29 May 2024.

Reshi, Ammaar (@ammaar) (2022d). “To the creators of these tools at OpenAI & MidJourney: how do we ensure protections for artists / train models on consent? Their talent, skill, hard work to get there needs to be respected. In fact we should involve them in the the creation of these tools so they’re heard.” 12 December, 5:20 p.m. Tweet, https://twitter.com/ammaar/status/1602337684046741511. Accessed 29 May 2024.

Reshi, Ammaar (@ammaar) (2022e). “This was all a test of what’s possible, but I’ll be waiting to hear more on the above before proceeding to create more content. Thank you to those who were willing to discuss despite the emotions this all unintentionally but very fairly spurred. I appreciate you.” 12 December, 5:20 p.m. Tweet, https://twitter.com/ammaar/status/1602337685971927045. Accessed 29 May 2024.

Reshi, Ammaar (2023). Alice and Sparkle. Independently published.

Schön, Finja (2023). Emily and the Mermaid Stone. Independently published.

Silva-Díaz, Maria Cecilia (2017). “Picturebooks and Metafiction.” The Routledge Companion to Picturebooks (Kümmerling-Meibauer, Bettina, ed.). New York: Routledge, pp. 69–81.

Stokel-Walker, Chris (2022). “A Tech Worker Is Selling A Children’s Book He Made Using AI. Professional Illustrators Are Pissed.” BuzzFeed, 13 December. Accessed 29 May 2024.

Stroh-Spijer, Tarwin (2023). “Creating a children’s book with AI.” Medium, 8 January, https://medium.com/maker-of-things/creating-a-childrens-book-with-ai-7c1dffa076d2. Accessed 29 May 2024.

Williams, J.P Anthony (2023). Lucy and the Enchanted Forest: Bedtime Story Book for Kids. Dream Weaver Tales.

Abstract

This case study examines the origins, content, and reception of the children’s book Alice and Sparkle, entirely generated by AI. Applying academic methodologies from the field of picturebook research, the authors reveal the internal incoherence of the work, resulting in, among other things, unintentional metafictionality. They analyse how the book was received – both online, where issues of authorship and ethics were raised, and among children, whose interpretations during a public school reading session aligned with insights yielded by scholarly tools. The text offers a pioneering, comprehensive analysis of a children’s book generated by AI – a phenomenon the authors suggest is likely to become increasingly common in the children’s publishing market.

Nikola Novaković – PhD, Assistant Professor at the Faculty of Teacher Education, University of Zagreb, Croatia, where he teaches courses in English literature. His research interests include humour in literature, particularly in children’s books, picturebooks, comic books, and graphic novels. He is the author of a book (in Croatian) on the function of humour in the works of Thomas Pynchon (2020).

Damjana Kolednjak – she is currently employed as a teacher at VI. Primary school in Varaždin, Croatia. She recently received her MA from the Faculty of Teacher Education, University of Zagreb, and her final thesis investigated the role of artificial intelligence in the production of books for children. Her fields of interest include children’s literature, picturebooks, and early English language acquisition.

The Polish translation of the paper (“Iskrzy czy nie iskrzy? O wygenerowanej przez AI Alice and Sparkle (2022) w kontekście badań nad książkami obrazkowymi”) was published in the book “Status sztuki dla dzieci. Pola, rozgrywki, refleksje [The Status of Art for Children: Fields, Games, Reflections], edited by Anna Czernow and Michalina Wesołowska (Poznań 2024). More information about the book (in Polish): https://biennaledladziecka.pl/rusza-sprzedaz-wysylkowa-ksiazki-status-sztuki-dla-dzieci-pola-rozgrywki-refleksje/

Polskie tłumaczenie artykułu („Iskrzy czy nie iskrzy? O wygenerowanej przez AI Alice and Sparkle (2022) w kontekście badań nad książkami obrazkowymi”) ukazało się w książce „Status sztuki dla dzieci. Pola, rozgrywki, refleksje” pod redakcją Anny Czernow i Michaliny Wesołowskiej (Poznań 2024). Więcej informacji: https://biennaledladziecka.pl/rusza-sprzedaz-wysylkowa-ksiazki-status-sztuki-dla-dzieci-pola-rozgrywki-refleksje/