Coexistence and creativity: screen media education in the age of artificial intelligence content generators

ABSTRACT

This article discusses the implications of Artificial Intelligence Content Generators (Gen-AI) for the field of screen media education. In light of the 2022–2023 releases of ChatGPT, DALL·E 2 and Midjourney AI, the article addresses the idea of what value there is for a student to enrol in a creative arts degree given the public view that Gen-AI threaten to replace human jobs in artistic fields. The article argues that there is potential for Gen-AI to offer major benefits to existing approaches to media education and that the technology could in fact drive greater interest in studying the creative arts. The act of creativity is valued by those who practise it for intrinsic purposes, and many graduates of screen programmes have traditionally leveraged their passion and creative conceptual understandings in a wide variety of employment fields outside vocationally-oriented screen production roles. Therefore, the article demonstrates that Gen-AI can benefit screen media programmes by improving employment opportunities for graduates, enhancing access and diversity for under-represented students, and can address the classic challenges of the theory-practice nexus for media production students.

KEYWORDS:

• Artificial intelligence
• screen media
• gen-AI
• storytelling
• digital media

Introduction

This article is being written as the release of ChatGPT, an Artificial Intelligence Content Generator (Gen-AI) created by OpenAI, has suddenly catapulted this type of technology into mainstream attention. Although it is not the first Gen-AI – nor even the first to produce generally convincing human-like text – ChatGPT captured the attention of news outlets due to its linguistic fluency. It also arrived against the background of 2022s breakthrough image generation platforms such as DALL·E 2 and Midjourney AI which allowed non-technical users to create digital artwork by typing descriptive prompts into a chat-like interface. For moving images, Google’s Imagen Video and Meta’s Make-A-Video have been demonstrated to take a text prompt (e.g. ‘Tiny plant sprout coming out of the ground’) and generate video (Ho et al. 2022). As of January 2023, the results of Imagen are remarkable although they have not yet attained the level of photorealism that can be achieved with still images produced by Gen-AI (Goldman 2022). The short videos that have been released – each lasting only a few seconds – have their own unique uncanny qualities which are visually interesting. It would almost be a shame for them to lose these qualities and become indistinguishable from genuine video recordings, although this is undoubtedly the aim of such research and development.

Future shock is easy. Toffler (1971) used this term to describe the effect of overwhelming experiences that impact ‘our perception of the future’ (Loloum 2020, 306). So-called ‘disruptive’ technologies such as AI and Virtual Reality have this tendency to immediately provoke the binary of evangelical and pessimistic discourse (Bender and Broderick 2021). But in this article, I want to put aside these matters and refocus the discussion on something more proximate: the purpose of studying a creative arts degree in this age of Gen-AI. In this time of a neoliberal push toward ‘vocational’ skills-based training in higher education (Di Leo 2013; England 2022), it is essential for educators to re-examine what they offer someone who wants to study screen/media production, photography, fine-art or even writing. Consider a potential student’s response to news headlines such as ‘Can Artificial Intelligence Replace Humans in Artistic Jobs?’ (Mosen 2022) or ‘Your Creativity Won’t Save Your Job from AI’ (Thompson 2022). However, I remain optimistic. If anything, I am more enthusiastic about the value of moving image education than ever before. First, I see the immediate potential for Gen-AI to offer major benefits to existing approaches in screen media courses. Second, and even more importantly, despite challenges that will accompany the existence of this technology, I can imagine it driving greater interest in people studying screen media (and other creative arts), thus positioning this area of education more centrally in the future of digital culture.

This article is concerned with exploring how and why Gen-AI could be incorporated into screen media programmes in higher education in positive ways. It builds upon my work in 2022 trialling various types of Gen-AI with students, and although the focus is on screen media specifically, the ideas connect to the conversation about creative arts education more broadly. This article is deliberately not evangelical, however, I also resist the urge to be pessimistic. My argument assumes that the act of creative arts practice is valued by those who practise it for intrinsic purposes (as it is for those who come to study various creative disciplines in our institutions). This of course recognises that the literature on creativity defines it as the process by which novel and new ideas and artefacts emerge which are valued and useful (Hennessey and Amabile 1998). I also acknowledge that creativity broadly speaking can include many human actions that are pleasurable and unrelated to screen arts production (Csikszentmihalyi 1975; Eisenberger and Shanock 2003). However this article is focused on the field of screen arts education. Against this background, as Daniel and Johnstone (2017) found in their analysis of student motivations for undertaking an arts degree:

While the entry points are slightly different including those returning to upgrade their skills or enact a career change, the majority of students (92%) were attracted to the notion of pursuing creativity, in fact for some of these (11%), there was simply no other option. (2017, 1025)

Therefore, the focus here is student motivation for studying (and completing) a creative arts course; rather than reduce options for graduate employment, I propose that Gen-AI has the potential to enhance their employment opportunities in a range of careers. I begin by summarising the background and general concerns about Gen-AI in creative arts education, including and especially the potential for the technology to threaten graduates’ jobs opportunities. I then outline some adjacent challenges faced by screen education such as the neoliberal push for vocational training in higher education, and pervasive issues associated with diversity. Finally, I present a clear argument for how Gen-AI can be leveraged by screen media education to improve inclusivity and ignite a renewed interest in theory and history within practical degree programmes. Ultimately, this is the beginning of a conversation designed to improve graduate employability from the perspective of enhancing students’ creative ‘mindsets’ by leveraging their creative ‘skillsets’ (Macdonald 2006).

Background

There are three types of Gen-AI this article is primarily concerned with. They are (i) text content generators such as ChatGPT, (ii) image-generators like DALL·E 2 and Midjourney AI, and (iii) video-generators such as Imagen. At the time of writing these are separate platforms, and Imagen is not yet available to the public. Nonetheless, the issues raised here are platform agnostic and apply regardless of which platforms become common, commercially available or open-source; or, for that matter, what type of hybrids emerge that can generate multiple modes of content. The basic user interaction with these platforms is that the AI utilises machine learning algorithms to generate content that is contextually relevant to a prompt entered by the user. For instance, ChatGPT might produce a story outline in response to a user prompt such as ‘Write me a 200 word story outline for a short film about a student struggling to pay their parking fine at college.’ The current image and video Gen-AI are based on a descriptive prompt about what the user would like to see in the artificially produced image. Here I quickly outline the key concerns that have appeared during 2022–2023, focusing only on those relevant to screen media education, and bracketing those common worries in other areas such as the medical fields, marketing and so on. These concerns have been drawn from the YouTube video interviews with experts, online discussions and news reports as well as a search for the terms ‘ChatGPT + jobs’ on ProQuest, which yielded 293 results of international news articles from December 16 to January 15.

After the release of DALL·E 2 and Midjourney AI, it became commonplace to read about their potential impact on the creative arts industry, with some arguing that they will disrupt the job market, while others positing that they would serve as valuable assistants to human artists (see Greene et al. 2023). This conversation exploded within two weeks of the release of ChatGPT in December 2022. It seems reasonable to suggest that Gen-AI has the potential to displace (some) human creative labour, and could lead to job losses in fields such as advertising, screen, and video game production (Salkowitz 2022; Villanova University 2017). This is a concern because of the already precarious nature of employment within the creative sector generally for recent graduates (Bridgstock 2019). Of course, in regards to screen education, these concerns are divorced from the reality that many students do not pursue creative arts degrees with the intention (or ultimate result) of attaining work in a specific role in creative field (Lewis and Lee 2020). As noted above, for many students the area is simply their passion (Daniel and Johnstone 2017).

Therefore, we should not expect the existence of Gen-AI to have a direct negative influence on student enrolments in the creative arts. Many commentators also noted the potential benefits that Gen-AI could offer as assistants, for example, increasing efficiency and enabling human artists to focus on more complex, conceptual tasks; in this sense, the technology could improve artist productivity (Darling 2022) as well as open up new possibilities for artistic expression (Clarke 2022). As I show in the remainder of this article, it is this latter position that offers one way of leveraging Gen-AI for the purpose of creative arts education.

While enrolments may not decline as a result of the existence of Gen-AI, the possibilities of student plagiarism and cheating on assignments are a potential concern. Of course, universities quickly realised they would need to embrace Gen-AI across all faculties and areas and recognised that they already had many systems in place for plagiarism-detection (Huang 2023; Illingworth 2023). Thus, academics began to consider approaching the situation by discussing in greater depth with students what constitutes academic integrity, while simultaneously committing to find ways to facilitate students’ use of Gen-AI in productive ways (Cassidy 2023). In addition, assessment in artistic fields has in any case already moved away from what Hanney (2013) described as ‘connoisseur assessment,’ toward an emphasis on process, often requiring elucidation of the creative process (evidenced in various ways) as well as including some form of reflection and/or exegetical material. As Hanney (2013) found: ‘The learning emphasis is on project completion and the delivery of a creative product of some sort, and the focus of assessment is on the students’ critical reflection on their own learning’ (53). Almost as quickly as the concern about cheating via Gen-AI appeared, those in higher education very quickly realised that the technologycould be of value to student learning and as the technology develops this article is intended to continue and stimulate further discussion in the field of creative arts education specifically.

It is also important to read the challenges (and opportunities) presented by Gen-AI within the landscape of contemporary higher education. One of the main issues is the continuing neoliberal shift, prioritising vocational training over more traditional forms of higher learning, particularly in the creative arts (Di Leo 2013; England 2022). Arguably, this is conspicuously inappropriate in the area of screen education where graduates do not typically move directly into a prescribed workplace ‘role’ for which their training has been directed. In Australia, for example, Bridgstock (2019) distinguishes between ‘business-to-business creative services,’ for example communications and marketing, and the ‘cultural production workforce’ which includes ‘artists, musicians and filmmakers’ (2019, 113). The former area has enjoyed economic growth whereas the latter has not. Even in neoliberal terms, a broad creative education is more likely to benefit the screen production workforce than narrow vocational training as it can enable graduates to pivot their ‘creative mindset’ (Macdonald 2006) toward a greater variety of ‘creative service’ roles. The sector also exhibits continued problems with diversity, with many groups underrepresented in production roles and in creative education (Dooley et al. 2022; Liddy 2022).

Discussion

The sorts of anxieties about Gen-AI in the field of screen media education are likely to be substantively different to those of other disciplines. After all, the typical creative arts student is unlikely to want to offload the practical components of their projects to an AI. I recognise that some students may wish to do so for all manner of reasons but, by and large, our students are motivated to enrol in these programmes because of an innate passion for the personal rewards associated with creating artworks, with substantially less interest in factors such as ‘social status’ or earning money (Daniel and Johnstone 2017, 1024). It seems doubtful that fine-art students, for instance, would prefer to churn out a series of artworks using DALL·E 2 and submit them for a final exhibition rather than produce the paintings themselves in the studio. Likewise, we would not expect the typical screenwriting student to prefer using ChatGPT to produce an entire screenplay and submit it for assessment. Informal polling of, and discussion with, undergraduate students through 2022–2023 has indicated a very low level of interest in simply ‘creating’ an artistic work of any type with AI. As an evolved human capacity (Dutton 2009), artistic creativity can be expected to continue to be a desirable activity for students.

Nonetheless, given the neoliberal context identified above, screen education will benefit from building a lively agenda that justifies its continued value in the age of Gen-AI. More importantly, this will position the field as stronger than ever before. Putting aside the obvious uses of Gen-AI to assist student essay writing and basic research, in this section my aim is to suggest three specific areas which I believe will serve as valuable starting points for screen education to productively integrate Gen-AI into student project/production work as well as the educational experience more generally. These are: (i) improving graduates’ employment opportunities by boosting their flexibility and refocusing attention on cultivating a creative mindset, (ii) improving options for accessibility, inclusivity and providing opportunities for media arts courses to act as broadening programmes for a greater range of students from other areas, and (iiii) emphasising the value of aesthetic judgement, critical and theoretical understandings in order to concretely bridge the theory-practice nexus.

Employment opportunities

The way screen education addresses the impact of Gen-AI on graduates’ jobs is crucial. Sam Altman, CEO of OpenAI, points out that the effect of AI on employment is shaping up differently than what was expected a decade ago (Greylock 2022). Instead of first replacing manual labour positions, followed by workers in fields such as medicine and law, and only finally displacing creative roles, the progression of AI is ‘going exactly [in] the other direction’ (Altman, interviewed in Greylock 2022). Thus, in the age of prompt-engineering an artwork, and as we watch Google’s Imagen create more and more convincing video from text input, what jobs would screen students expect to move into as they graduate?

The simplest and obvious way in which integrating Gen-AI into screen programmes is that we can safely assume graduates will be working with AI tools in whichever industry they move into. Therefore, this essential digital literacy will involve ‘co-composing’ (McKnight 2021) written texts, but already AI assistance has found its way into some rudimentary visual effects tasks, as well as other editing tools. There are clearly larger opportunities than simply equipping students with the skills to operate Gen-AI. It is essential to remember that employment for creative students has long been identified as one where graduates need to pivot their skills and understandings into other fields that will generate income. Bridgstock and Cunningham (2016) describe this accordingly:

Moving from creative education to work is more like ‘translation’ rather than transition, with graduates needing to recontextualise and reinterpret knowledge, capabilities and practices acquired during degree courses for an extremely wide variety of employment situations, a process that often occurs alongside significant and fundamental shifts in career identity’. (2016, 21)

This process of adjusting one’s career identity aligns with Shine and Cullen’s (2019) position that we should ‘educate for employability (growing abilities) rather than employment (securing a job)’ (2019, 244). Bennett (2009, 311) calls these capacities ‘protean,’ and suggests that successful graduates from creative degrees have the ability to constantly adapt and learn new skills in order to take on new opportunities. As I show below, the incorporation of Gen-AI into screen media courses is one way in which educators can develop more strategies and opportunities to develop such a protean approach as part of the larger project in screen education to reconcile the theory-practice nexus (see Wright-Brough et al. 2023). Or, as de Jong (2006) notes, taking students ‘from "doing" to "knowing what you are doing"’ (2006, 151).

These ideas are extremely well illustrated in a conversation with Brendan Seals, visual effects supervisor at Luma Pictures who noted that the amount of self-learning in post-production that has been available online in recent years means that when employing entry-level visual effects artists it is no longer enough to simply be able to use some particular piece of software. Rather, employers are looking for how someone can apply concepts, what capacity they have to generate ‘original ideas’ and bring imagination to their execution (Seals, in The Future Of 2022). Thus, Gen-AI will need to be incorporated in ways that enhance these capacities and flexibility. After all, according to one study in which creative arts graduates were asked how they felt that they ‘added creative value in jobs’ that were outside their direct discipline, almost two thirds indicating this contribution was ‘Through providing creative viewpoints and ideas (33%) [and] through possessing critical thinking skills (31%)’ (Nielsen, Bridgstock, and McDonald 2018, 167). This notion of enhancing employability by cultivating a creative mindset instead of being mired in focusing on students developing skillsets in particular roles (Macdonald 2006) is the concept upon which the following sections hinge.

Access, diversity and broadening opportunities

Gen-AI can provide access opportunities for people who have wanted to be able to undertake creative arts study, but have been historically misrepresented (regardless of their creative potential) due to demographic and social inequalities. This issue remains timely; one report by the Australian Screen Production Education & Research Association found that ‘while similar numbers of male and female students are completing capstone projects in tertiary (higher education) screen production departments and/or film schools in Australia, crew roles [remain] highly gendered’ (Dooley et al. 2022, 120). At the same time, Liddy’s (2022) research into industry professionals’ experience of diversity in the media industry discovers there do not seem to be ‘any seismic shifts in the screen industries, but most [practitioners] acknowledge that there is greater awareness of the importance of gender equality and diversity and some (welcome) change has occurred during the period 2016–2021’ (2022, 24). Therefore, the moment of Gen-AI can be leveraged to explore ways to improve diversity, access and open up the media production curriculum in innovative ways.

This aligns with a number of recent shifts in creative arts education generally, which is an area that has long recognised the importance of ‘accommodating learning styles [in] writing learning outcomes and assessment activity’ (Appleton, Montero, and Jones 2017, 149). For instance, many educators already make accommodations such as offering students the opportunity to produce video essays rather than written essays (Redmond and Tai 2021). Against such a background, this is one obvious way in which screen education can incorporate the use of text Gen-AI as ‘co-composition’ assistants (McKnight 2021) for students in assignments where they need to write any type of analysis of a screenwork. Students could co-compose a written essay with a Gen-AI, or they could use it to assist with the writing of the verbal script to be used as voice-over in their video-essay. By leveraging the rapid-writing afforded by co-composing with an Gen-AI, students would be able to focus on the development of their ‘critical literacy and creative imagination,’ which are more authentic characteristics of creative education than focusing on the mechanics of a written essay (Redmond and Tai 2021, 8). There is obviously an employability capability here, with one student’s reflection on producing a video essay commenting: ‘if I were to be going for a job interview somewhere, I would probably whip out one of the videos that I did for university or a physical thing rather than just shoving an essay in their face’ (2021, 14). Thus, in line with earlier points about improving graduates’ employability, the opportunities to make screen education more inclusive for a range of traditionally underrepresented groups is of clear benefit to the education and creative sectors, as well as adjacent employment fields more generally.

There are significant opportunities for cross-disciplinary student work both in terms of formally arranged projects as well as by simply encouraging greater enrolments in screen media classes for students from fields and disciplines outside of the creative arts. Gen-AI may have arrived at the right time; universities are now far more friendly to interdisciplinary approaches than even two decades ago, at a time when Lindauer suggested that ‘education excursions across disciplinary boundaries are rare’ (Lindauer 1998, 2). For example, a diverse range of other fields, including engineering and social-work, have been incorporating creative opportunities for assessment such as video-essays (Caratozzolo et al. 2022). Gen-AI may thus provide a ‘way in’ to arts and creativity for students from other courses wanting to broaden their skills, as well as people who have felt like they may not have the talent to study but nonetheless have a creative mindset and passion. In addition, more formalised kinds of cross-disciplinary projects have been shown to improve student outcomes in their primary area of study (Cattoni et al. 2022; Dowds 1998; Ryan et al. 2023). For example, a nursing student may not need to wrestle with the task of producing a shot-list for their creative project if they can instead speed up the process with the help of Gen-AI. This would enable them to focus on the critical outcomes of the project such as developing character in order to cultivate (in themselves) a greater sense of empathy, arguably an essential skill for their career (Bas-Sarmiento et al. 2020). It is also hard to overestimate the reciprocal benefits to broadening opportunities for students from other areas and introducing them to media production ideas, techniques and processes: screen students will benefit from the modelling of developing transferable skills, which are ‘of particular significance in these fields, where students will generally find employment outside their subject areas’ (Cox 2012; Walmsley 2013, 227).

Theory-practice nexus

A long standing issue in the practical education of film and screen media is facilitating student understanding (and appreciation) of the fundamental connections between the practice of making a screenwork and the rich history of aesthetic, critical and theoretical scholarship of cinema and media studies (Petric 1976; Tomasulo 1997). Consider the following statement from Jillian Holt, reflecting on the difficulties associated with teaching students to learn the art, craft and technique of editing: ‘How can creativity in editing can be taught when so much of what editors do is attributed to “intuition” or “what feels right”?’ (Holt 2018, 179). This embodied nature of intuition (Bastick 2003), in the field of screen media production, can be developed through experience and reflection on the results of one’s creative decisions (Bender and Sung 2020). Intuitions are also cultivated by actively working on a creative mindset with the capacity to harmonise and go beyond existing norms of style from a historical appreciation of diverse range of artworks (Gombrich 1995). According to confluence theory, intuition is a key component of creative thinking and learning along the way to developing insight, for instance ‘ordering, connecting, and selecting ideas [through a process of] personal impact and intuition ("Does it touch me? I feel there is something in it")' (Hoorn 2014, 114). As such, this process also includes understanding the contributions of critical theory to our understanding of screen productions and audience reception (e.g. genre studies, basic Formalism, neorealism(s), the ‘gaze’ etc.). These traditions are part of the discipline’s rich intellectual history and should be first and foremost in the curriculum, but so too should be the insights of poetics (Bordwell 2012). Thus, an integration of theory and practice may involve students utilising theoretical concepts within their productions; Tomasulo’s examples indicate how the ‘gaze’ and Bazin’s conceptualisation of long-take cinema find their way into student productions in fairly explicit stylistic expressions (Tomasulo 1997).

Against this background, which of course makes sense to educators, students frequently struggle to understand how the theory-practice nexus is useful and essential (Wright-Brough et al. 2023). I contend that Gen-AI offer a unique problem (and solution) in this context. To attain the desired outcome with Gen-AI, it is essential for the user to have a strong ability to articulate their intention clearly (in addition to simply having a creative idea). As Sam Altman, CEO of OpenAI, pointedly states: ‘What will matter [in the future is] the quality of ideas and the understanding of what you want’ (Greylock 2022). Furthermore, Altman emphasises that the most successful generation of AI content will be achieved when the artist/user has a clear ‘vision’ of what output (or creative direction) they want to accomplish with Gen-AI. This can be demonstrated to students, and indeed students can discover it for themselves, through a structured task in which they use Gen-AI to generate ideas. An example of this follows in the next section on articulation of intention, but in general terms it is easy to imagine a classroom activity where students are provided with a list of prompts that gradually escalate in their clarity of articulated vision, for instance:

1. Write a 300-word character outline for a sci-fi film set in space;

2. Write a 300-word character outline for a dystopian sci-fi film set in the year 2075 and that takes place in outer space;

3. Write a 300-word character outline for a female astronaut in a dystopian sci-fi film set in the year 2075 and that takes place in outer space;

4. Write a 300-word character outline for a female astronaut in a dystopian sci-fi film set in the year 2075 and that takes place in outer space – the major dramatic conflict is internal, she is undergoing a crisis of faith about whether she can fulfil the requirements of her mission.

Students will find that the content generated by the more detailed prompts is not only more sensible but likely to be more interesting. A brief class discussion of the experience of this activity will reveal that Gen-AI works best for people who think creatively, but who also use this mindset in an applied way (e.g. what is the conflict this character is facing?). Altman’s claim implies that someone who ‘knows’ art history will be able to produce something of far higher quality; this obviously requires an understanding of arts history and theory, as well as an appreciation of how audiences respond to specific artistic techniques. Regardless of whatever task(s) the students use Gen-AI for subsequently, a simple activity like this can make very clear how essential it is to engage meaningfully with history, theory and criticism of screen media in order to improve their own creative capacities. There are no doubt many more ways to incorporate Gen-AI into cultivating students’ abilities (and appreciation) of the theory-practice nexus, and while this is a simple starting point the accessibility of these platforms represents a unique opportunity to experiment and push students in innovative ways.

Articulation and intention

I believe that the three areas of integration identified above (boosting employment opportunities, improving access and delivering the theory-practice nexus) ultimately come together around students using Gen-AI for the purpose of improving their capacities to articulate their creative intentions. The unfortunate caricature of using Gen-AI for ideation might be a student simply churning out ideas for films rather than undertaking some form of structured or free-form ideation process to develop what they want to create. However, both text-based and image-based Gen-AI are without doubt useful tools to assist in the ideation process. It can often be difficult for people in general (and students are no exception) to articulate their visual thoughts (Costantino 2007; Efland 2004; Zeki 1999), which is obviously problematic in creative production projects. Thus, Gen-AI can help students to rapidly mock-up prototypes of visual ideas including colour palettes, framing and composition, potentially salient objects, mood and lighting etc.

In this section I illustrate the concept with a simple example: let us imagine two hypothetical groups of students using an image-oriented-Gen-AI (e.g. DALL·E 2) to undertake an ideation activity. Group A consists of students who have developed good skills of articulation and already have some intentions for their project. Group B has neither. Of course my two groups are exaggerations for the purpose of demonstrating the potential for Gen-AI, and readers will be able to apply the type of thinking to other real-life examples to use in the classroom. For the purpose of this exercise, the students are expected to develop a short film concept and they will use DALL·E 2 to rapidly produce iterations of various ideas related to a particular theme. The utility of this is that current generation image-output Gen-AI do not generally produce any kind of ‘finished’ output; rather they produce an initial set of rough first drafts that represent somewhat different AI ‘responses’ to the prompt. The user can then refine one or more of these images through a series of iterations, in a sense pushing and guiding the AI in the creative direction the user wants (illustrated in the iterations from Figures 1–3 below).

Figure 1. Group A’s first prompt to generate ideas about potential mood, colour and light.

Figure 2. Group A’s second iteration; adjusting the mood.

Figure 3. Group A’s third iteration; adjusting character and retaining mood.

Throughout this demonstration, it will be clear that the process is an example of Sawyer’s (2022) idea of using ‘external representations – such as graphs, figures, maps, and texts, either on paper or a computer screen’ to ‘enhance learning’ by ‘fostering metacognition and collaboration [facilitating] new methods of scaffolding and formative assessment’ (2022, 459–460). This task will be an exploratory process for the groups; let us imagine that both groups happen to have chosen themes associated with a dystopian film set in the Australian outback during an energy crisis. Their next step is to consider the general mood of the project and perhaps determine their approach to how colour and light will be used to evoke the themes. Group A uses the following prompt to produce their first iteration: ‘photorealistic image, f2.8, evening sunset over dystopian landscape, outback australia, a wind turbine in the foreground is broken, rusty; a man in dirty military uniform sweats and stares into the distance, the man is in close-up wiping sweat from his face’ (Figure 1). They decide these images convey the appropriate iconography and perhaps even characterisation, but the lighting and mood are overly dramatic and do not quite communicate a dystopian setting. So their second iteration uses only a minor change to their prompt, replacing ‘evening sunset’ with ‘middle of the day, gray sky’ (Figure 2).

Now, satisfied with the mood of iteration 2 (Figure 2), the group decides they want to try their main character as not associated with the military. So they change that aspect of the prompt to ‘man in dirty overalls’ as substitution for ‘man in dirty military uniform’ (Figure 3). This process could be completed in about ten minutes. Arguably, this is really just using Gen-AI to speed up the process of constructing a mood board from reference images. Yet, in only ten minutes with Gen-AI, the group can be much closer to a shared understanding of the project’s intentions. They would now be in a position to construct a more traditional mood board, and that pre-production process would not only be faster but would have a more complete sense of direction.

Now contrast this to the work of Group B (Figure 4). Group B does not seem to have the pre-existing knowledge (or skills) to articulate the mood desired, understand that the time of day would affect the lighting in relation to this mood, or that the mise-en-scene of body language and character costuming will contribute to the drama. Their prompt was created by taking the theme of ‘dystopian future due to energy crisis’ at a basic level instead of considering the narrative elements that might evoke a sense of such a dystopian world. As such, they created the simplistic prompt: ‘photorealistic image, dystopian australian outback drama about energy crisis.’ Obviously, each of these images produce by DALL·E 2 is really interesting! But they would be far less helpful for the student group to move to the next phase of pre-production.

Figure 4. Simplistic understanding of how to use the software results in interesting and exotic, but irrelevant imagery.

Consider the taxonomy of photographic, cinematic, thematic and conceptualisation skills that can be demonstrated here. In a sense, this task could even function as a kind of formative assessment to evaluate students’ (or groups’) understanding of particular concepts and skills. Further, scaffolding could be provided for students in Group B to move toward the type of understandings demonstrated by Group A. But more importantly, this exercise illustrates that the development of articulation skills needed to use Gen-AI effectively provides a context for everything else that already happens in media production courses.

Building upon this simple example, we can see that a structured encounter with Gen-AI could also provide access for students who may learn differently; take the simple example of the effect of aperture on an image. It can be difficult for some students to understand the effect of aperture by simply seeing examples in a classroom powerpoint, general discussion or textbook. Working practically with a camera is an ideal way to learn and understand the impact of different f-stops on the depth of field, however for the novice it can also be easy to become distracted by all of the other technical issues of setting up lights, adjusting neutral density filters etc. Simply altering the prompt in an image-oriented Gen-AI can yield a variety of images illustrating the basic differences between f2.8 and f11.

This example illustrates three immediate opportunities for creative media education:

1. It demonstrates that in order to use this technology effectively, students will need to (continue to) be trained in higher-order conceptual and theoretical understandings of their artform;

2. It would also be possible to use an activity like this for either formative or summative assessment purposes; e.g. assessment could be focused on the prompts students provide, rather than necessarily the image produced as an output;

3. Such activities can help motivate students to refine their articulation skills by rehearsing with Gen-AI and reflecting on the process.

This does not mean that every single student production would incorporate Gen-AI. Their use may be as simple as incorporating them into an idea generation session during class as demonstrated above, or it may be as extensive as allowing students to co-compose an entire script. At the moment, Gen-AI-produced scripts tend to contain limited subtext and convey ideas through extensive (and obvious) dialogue, however, these features are also typical of the work of novice human screenwriters (Caplin 2020). Perhaps students could get more quickly to the same quality first-draft with Gen-AI assistance, and then the real work would be in refining it, becoming judicious about which lines of dialogue are necessary and where ideas can be better dramatised as subtext etc. Therefore, although this hypothetical classroom activity demonstrates only one basic model of using Gen-AI in the ideation phrase, many more activities and integrations will be discovered by the field as students and academics incorporate Gen-AI more readily into class, assessment and project work.

Conclusion

Artificial Intelligence Content Generators (Gen-AI) pose many interesting dilemmas for creative arts education and for screen media academics specifically. Research will explore how Gen-AI impacts the ontology of the photographic and cinematic image, no doubt continuing and extending previous work on digital post-production (Prince 1996, 2011) and DeepFakes (Pan 2021). In addition, the issue of originality will be addressed and extend earlier work beyond remix culture and hypermediation (Bolter and Grusin 2000). An immediate step of course is to address how AI impacts the ‘aura’ of art, as per Benjamin’s ([1935] 1969) analysis of the age of mechanical reproduction (and from which this article takes its title). There is also the significant issue of IP ownership and copyright, particularly in relation to the presently unresolved ethical concerns around how the image training datasets have been (and continue to be) harvested from artworks without proper attribution or consideration of artist/copyright holder royalties (see, for instance: https://haveibeentrained.com/). More proximate for the education sector however, Gen-AI presents students with the ability to rapidly produce content that seems convincing, and could be submitted for assessment. Against this background, this article has recognised that it is understandable for many areas of higher education to become fiercely introspective about how to manage Gen-AI: whether to ban their use, educate students in their potential limitations, or to embrace them and find ways to incorporate Gen-AI into student work.

But screen media and other creative arts education can be viewed as a unique milieu in which to consider the impact of Gen-AI. After all, our entire field is about generating content. Our students come to study creative degrees in order to attain careers ‘generating’ content. After ChatGPT, DALL·E 2, Midjourney AI and Imagen, we have now arrived at a point of coexistence between human and machine creativity. The fact that there is still value in a creative education throws into sharp relief any neoliberal expectations that the area should focus on vocational training (Hennessey 2022). This article has argued that Gen-AI presents a unique opportunity to re-investigate the divide between ‘creative skillsets’ and ‘creative mindsets’ (Macdonald 2006). First, I have shown that Gen-AI offers multiple opportunities to improve graduates’ employment opportunities by developing their industry-ready digital literacy capabilities to work with AI tools. This also refocuses student attention and learning on the essential transferable conceptual skills associated with creativity (Eisenberger and Shanock 2003; Sawyer 2021). Second, Gen-AI offers the capability to improve accessibility and diversity for students that want to study screen media, as well as generating occasions for the area to function as an effective broadening subject for students from other areas. And third, there is a capacity for Gen-AI to aid in bridging the theory-practice nexus that has long challenged students; this will benefit students by cultivating greater aesthetic taste and critical visual literacy, and also the ability to apply these conceptual understandings to their own practical work. Finally, the article demonstrated one very simple model for integrating Gen-AI into a screen production ‘ideation’ session for students that is designed to enhance student’s capacities to conceive and articulate their creative intentions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Stuart Marshall Bender

Stuart Marshall Bender is Associate Professor in Screen Arts at Curtin University. He is a filmmaker with expertise in digital visual effects and teaches screen production and theory in the context of a creative arts degree. Stuart's research specialty is understanding the impact of high-emotion media content on audiences.