AI in Visually Based Creative Industries: What Early Research Tells Us About a Changing Creative Landscape

AI entered an established creative ecosystem

One of the important distinctions made by the researchers is that AI itself was already established within digital visual production. What changed around 2022 was the accessibility and quality of generative systems such as DALL·E, Stable Diffusion and Midjourney.

For the authors, this represented a qualitative shift. Generative tools had reached a point where their outputs could begin influencing established production practices and, potentially, the livelihoods of professional visual artists.

The paper also highlights the relationship between generative AI and the enormous datasets required to train these systems. The ability to generate an image from a text prompt depends upon pre-existing visual material, including photography, painting, film imagery, game assets and CGI. This relationship between existing creative work and newly generated content sits at the centre of many of the legal and ethical questions explored throughout the research.

Creative control remains a significant part of the equation

The researchers are particularly interested in the difference between generating an image and directing a creative outcome.

Early generative systems could rapidly produce visually compelling material and unexpected creative possibilities. However, the researchers found that achieving a specific result could be considerably more difficult. When a practitioner had a precise visual intention, repeated prompting and iteration could become frustrating and time-consuming.

The authors also observed that generative systems tended towards aesthetic and compositional patterns contained within their training data. This raises a broader creative question that remains highly relevant: if a system is particularly effective at producing variations based upon established visual conventions, what role does the practitioner play in challenging those conventions?

Within professional creative production, speed of generation is only one measure of usefulness. Consistency, accuracy, intentionality, authorship and the ability to respond to a wider creative brief all remain important parts of the production process.

Different creative professions face different forms of change

Perhaps the most interesting part of the research is its refusal to treat the creative industries as a single category of work. The authors instead consider how AI might affect particular specialisms differently. They predicted considerable disruption for some forms of commercial illustration and concept art, particularly where clients require relatively contained, one-off visual assets. At the same time, they anticipated greater resilience in work requiring sustained visual consistency, specialist knowledge or significant creative direction.

Animation provides a useful example. The researchers predicted that highly repetitive processes, including colouring, in-betweening and lip-syncing, could increasingly become automated, while roles requiring substantial creative input would remain dependent upon human practitioners and supervision.

Similarly, they argued that storyboarding and comics presented challenges for generative systems because these formats depend upon consistency across characters, environments, props and sequences of images. At the same time, they recognised that generative tools could make visual development more accessible to independent and lower-budget creators who previously lacked the resources or drawing skills to explore their ideas visually.

This distinction is important. AI adoption does not necessarily produce the same outcome across every creative discipline. Its impact depends heavily upon the nature of the task, the level of creative judgement involved and the standards required of the final work.

Automation can also expand who gets to create

The paper also raises an important point that can be lost when discussions focus exclusively on job displacement. An increase in AI-generated visual content does not automatically represent an equivalent reduction in paid creative work.

The researchers point towards individuals and small organisations using generative tools to produce presentation graphics, simple visual identities, website imagery and other materials that they may never have commissioned professionally.

Generative AI can therefore expand access to forms of visual communication that previously required greater technical skill or financial resources.

This creates a more complicated picture of AI’s economic impact. Some established creative tasks may face increased automation or pricing pressure, while entirely new forms of participation, experimentation and creative production become possible.

New tools create new forms of creative expertise

The researchers also anticipated the emergence of new professional capabilities around AI itself. Even in 2023, job advertisements were beginning to appear for roles such as “AI Artist” and “AI Concept Artist”. The authors predicted that familiarity with AI-supported design methods would increasingly become part of the broader toolkit expected of digital creators.

Their observation points towards a wider pattern in technological change. Creative roles rarely remain completely static when production technologies change. Instead, the boundaries between roles, tools and skills are continually renegotiated.

For emerging creative professionals, this places increasing value on understanding the wider production process. Knowing how to operate a particular tool has value, but so does understanding where that tool belongs within a workflow, what constitutes a successful creative outcome, and when its use introduces additional ethical, legal or practical problems.

Copyright and ownership sit at the centre of adoption

The research gives considerable attention to copyright, authorship and ownership. At the time of publication, major questions surrounding AI training datasets and the copyright status of generated works were unresolved. The researchers recognised that decisions made through legislation and legal cases could substantially influence both the development of generative technologies and the willingness of professional organisations to incorporate them into production.

The paper highlights an important tension between technological development and the rights of creative practitioners whose work contributes to the datasets upon which many generative systems depend.

For professional creative industries, these questions extend beyond abstract legal debate. Ownership, provenance and the ability to control intellectual property can directly affect whether an asset is suitable for commercial production.

These considerations also reinforce the importance of AI literacy extending beyond technical operation. Creative professionals increasingly need an understanding of the ethical and legal context surrounding the technologies they use.

What does this mean for creative education?

The paper concludes by recommending that digital artists familiarise themselves with AI systems and explore how these tools might supplement their existing practice. Crucially, the authors also argue that automated creative processes still require oversight from someone who understands the underlying creative process. This has significant implications for creative education.

Students entering animation, design, film, games and related visual industries will encounter production environments in which AI capabilities continue to change. Preparing them for that environment requires more than familiarity with whichever platforms are currently dominant.

Creative fundamentals become particularly important when technologies can produce outputs rapidly. Practitioners still need to recognise whether an image communicates effectively, whether a design supports the wider creative intention, whether visual continuity has been maintained, and whether an output is appropriate for professional use.

The ability to make those decisions comes from disciplinary knowledge, visual literacy, communication, critical thinking and experience of the wider creative process.

AI literacy therefore needs to sit alongside, rather than displace, these foundations.

Human oversight remains central to creative production

The technologies discussed in this 2023 research have already developed substantially. Some limitations identified by the authors have diminished, new capabilities have emerged, and the legal and professional environment surrounding generative AI continues to evolve.

Yet the broader questions raised by the paper remain highly relevant. Which tasks can be accelerated? Which skills become more valuable when production becomes increasingly automated? How should creators protect authorship and intellectual property? Where should responsibility sit when AI becomes embedded within a professional workflow? Perhaps the most useful conclusion from the research is the authors’ insistence on understanding the creative process itself.

As individual tools change, creative professionals still need to understand what they are making, why they are making it, and how individual production decisions contribute to a larger creative outcome. For educators and industry alike, that may prove to be one of the most durable capabilities we can develop as AI continues to reshape visual creative practice.