Cynthia Kingsley-Smith
Illustration
→ Portfolio
→ Creative CV
→ Free-to-use Textures
→ About // Contact
Thematic Essay - Reflective Practice
AI Ethics and The Importance of Human Design
Cynthia Kingsley-Smith
Artificial Intelligence has become a key component in the growth of various industries such as the tech, finance, and medical sectors. The mass adoption of AI in both complex problem-solving within the sciences, and for automating more simple day-to-day tasks is in some ways inevitable and generally accepted as an OK thing. However, the advent of generative AI as a process through which visual art can be created has caused a massive issue for the illustration and creative arts industry. This essay will use contemporary examples and data points where available to look at how AI has disrupted the art industries, why AI is considered to be indicative of low-quality work which lacks authenticity and provide practical and theoretical arguments for why the preservation of handmade art and design is essential to preserve human ethics, creativity and cultural works.
It is correct to describe AI art as “art”. Any method an individual may use to express their unique vision on a topic or thing could likely be described as art in a literal sense – the issues I will discuss with this methodology are rooted in an incredibly shaky legality, a lack of ethics, a lack of technical or general understanding of how this software works and ultimately, a disregard for how this software adversely affects artists living in the real world today, as well as the users capacity to interface and connect with a visual artistic experience.
The very way large-scale generative AI models exist today necessitates the theft of copyrighted artwork. A “model” such as Midjourney, DALL-E or Stable Diffusion ran by a private company will be given a data set such as LAION 5B, which is collection of 5.85 billion images and URLS that are paired with descriptive text, with which the software model will run a complex mathematic formula in order to assign expected visuals to certain words. The input of this data set is the absolute in determining what the model can output. LAION datasets are commonly used because they are curated by a German non-profit company for research purposes, which completely shields the organization from any legal issues that could arise from harvesting copyrighted data. This means that AI companies are openly utilising charitable legal exemptions to feed data into for profit ventures in a manner that is barely legal. OpenAI even openly describe how they will invent a new legal structure to support their capital gains utilising this loophole in their own 2019 blog post:
“...no pre-existing legal structure we know of strikes the right balance. Our solution is to create OpenAI LP as a hybrid of a for-profit and nonprofit—which we are calling a “capped-profit” company”.
One important distinction to note is that the anti-AI campaigners within the music industry have largely had a more immediate and effective response to things like Spotify hosting thousands of AI-generated albums and the non-consensual use of their music in training AI models. Thanks to consultation from music industry heads and mass communication from members of the public, the UK government released a Report on Copyright and Artificial Intelligence in March 2026 in which they backtracked on their previous stance which allowed AI companies to use all copyrighted works to train models as long as there was an “opt-out” option. While it is worrying that the influence and star power of the music industry was required to get the UK government body to reconsider their stance, this is still a massive victory for all creatives as using an opt-out solution is an inherently predatory practice. These systems do not protect the copyright holder from the versions of the AI that already exist and have been trained on illegally used data, and relies on the artist being both aware of the option and tech-savvy enough to understand the process to opt-out, which may even require giving additional personal information to another business.
The rise of generative AI has underpinned a massive loss of job opportunities across all creative sectors, and especially for freelancers. The British Design and Artists Copyright Society released a report in 2024 finding that earnings for visual artists in the UK have fallen by 40% since 2010. This is likely to get worse as employers will be able to leverage AI tools to replace mid and entry-level artists and offer lower salaries to artists who are willing to use the technology to improve the speed of their workflow, even if they ethically disagree with it. This will disproportionately affect the barrier to entry for artists looking for entry-level or lower-paid jobs.
Contemporary art galleries have already hosted collections dedicated entirely to art created with generative AI and sold pieces worth tens of thousands of pounds. As consumer markets are flooded with more AI generated artwork which is capable of responding to user requests and rising trends in a matter of minutes, it will force us as creatives and as a society to collectively re-examine what “valuable” art is. The argument that generative AI makes art “more accessible” also loses its credibility when a user realises that outputs from an AI model will lose value because there is no scarcity nor effort in product – the generations will simply join other tens of thousands of similar outputs in the vast online sea of data. I believe in response to this; more value will be placed on the story behind a piece of art, and the process that made it.
Above: Taken from the 20 Feb – 5 Mar 2025 Augmented Intelligence auction, https://onlineonly.christies.com/s/augmented-intelligence/lots/3837.
A point in history that is extremely similar to the social shift we are seeing today is the rise of the Arts and Crafts movement in the late 19th century which occurred as a response to a large boom of industrialization and subsequent perceived decline in standards in both art and consumer goods. Aside from the highly decorative aesthetic, the movement can be identified by its adherence to specific philosophies that are particularly relevant today. Ideas like “truth to materials” can refer to artistic integrity and the idea that an artist should be capable of understanding and utilizing their mediums to the best of their natural ability. One quote from William Morris made in 1884 tackles the ways in which many creatives find generative AI uncomfortable today.
“Nothing should be made by man’s labour which is not worth making, or which must be made by labour degrading to the makers.”
I think this quote effectively puts forth that production of any given thing should both be worth doing and doing in a manner that is ethical, which are both things that generative AI fail at through the complete removal of any effort from the artist and use of stolen content to make it happen.
Above: Trellis Wallpaper, William Morris, 1862. https://www.wmgallery.org.uk/object/design-for-trellis-wallpaper/
- “AI is to be embraced by artists as a collaborative partner, rather than a replacement”
Treating the rise of generative AI tools the same as any other new invention is problematic as it totally dismisses the issues already caused by the damaging language used by AI companies when initially advertising these products. The largest portion of the userbase for these products are people with little to no artistic knowledge or experience, who have neither the skill nor desire to “touch up” or continue working on an image after it is generated. We can see that companies are aware of this from the language used in advertising and version release blog posts; the artists nor artistic techniques are very rarely mentioned, and the effectiveness of software is measured by how little editing the image needs pre- or post-processing.
One particularly relevant study by Paolo Cardini, Thode Island School of Design and Juan Noguera, Rochester Institute of Technology examines an experiment where Guatemalan Traditional Craftspeople are given commissions to create bronze door knockers, with the caveat that they use DALL-E image generation software to generate and refine their ideas in the concepting stage.
After talking about how the craft market demand is small, the conclusion of this study posits “It is financially safer for the artisans to give their audience what they want, securing their profits at the expense of their creative freedom and evolutionary push”. I find this suggestion that the opportunity for these craftsmen to work with AI has boosted the inherent “creativity” is biased and reductive in that it totally disregards the current expertise of the craftsmen without giving them the opportunity to tackle a similar problem without the use of AI.
- “Artificial Intelligence just collects inspiration and copies from references exactly as artists do”
No matter what, if an experienced human artist attempts to copy the work of another, there will be traces of that individual artists' experiences and influences within the work. It is in this amalgamation of experiences and knowledge that iterations on work become something truly human and unique. Even from an economic perspective, this is an essential element in how artists are able to contribute to the cultural output of a country. AI is absolutely incapable of this process and will simply copy the given image verbatim. An open university of Catalonia research paper suggests that AI, as a tool, can automate 26% of tasks in the arts and design sectors. I believe that by offloading these “soft skills” tasks to AI we are likely to see catastrophic effects on art and cultural development of a country and deprive human artists of what can be a personally enriching and enjoyable part of their job.
This also alludes to another issue with datasets in generative AI – the bias of data. As AI models simply use the images they are given and do not make mood boards nor consider their choices of reference nor understand the cultural significance or deeper meaning behind any given image, they are very prone to participate in harmful, exaggerated stereotype. This is especially dangerous as the mechanical nature of these outputs may work to further legitimize these stereotypes for any business using the software, such as an advertising agency or law enforcement.
Above: Bloomberg Analysis of Stable Diffusion, Leonardo Nicoletti and Diana Bass, https://www.bloomberg.com/graphics/2023-generative-ai-bias/.
In addition, iterative data sets suffer from an ouroboros-like homogenization. As these images become more common generative AI models are fed more AI generated images to reference, which causes unexpected results and pushes the “uncanny valley” feeling of AI imagery up to 11, whist still reinforcing the inherent biases that AI is unable to reason with. Overall, this is another important reason skilled artists and craftspeople are essential figureheads in creative industry and should resist being replaced by AI. As artists, we can content with these complex issues and ideas such as systemic bias and by accounting for them in our work through human empathy, consideration and proper research we can combat such problems rather than reinforce them.
Unfortunately, artists cannot reclaim their stolen artworks, and even on an individual level it is impossible to know the true extent of financial, reputational and emotional damage done to those who have had their unique artist handwriting stolen and used specifically in data sets to be as easy to replicate with AI as possible. Notable examples of this are effects on legendary concept artists Lois van Baarle and Greg Rutowski.
Above: Example of an illegally trained “style” model taken from Google SERPs, 2026.
It is entirely possible to have such a version of generative AI software that is ethical, one that can leverage the more genuinely useful aspects of generative AI such as the ability for a video game to save time during the prototyping stage by using more accurate placeholder assets. If the dataset used was made up of public domain images, and then artists were compensated lump sums or royalties for the use of their work in an “opt-in” system, then there would be much less resistance to this new technology. I think this method of adopting an ethical approach to AI development would naturally redefine the software as a more collaborative tool instead of a replacement, and bring the technology more in line with other developments in the world of illustrative and creative technology such as how some artists learned to render images with Blender, or how the invention of the Camera reduced the amount of artists who learned photorealistic draftsmanship. There are already third-party certification services which ensure that a given artificial intelligence model is not trained on any stolen data, which I predict will become vital in winning back consumer trust that product you are selling is not low effort despite using AI.
Above: Footer website certificate regarding fair use of copyrighted assets. fairlytrained.org
Another valuable step that creatives in society can take to help stem the cultural bleeding caused by AI is to consider a slightly different approach in art education. I believe that understanding your copyright rights as an artist not just in terms of client contracts, but also in terms of how to defend your copyright against more nebulous threats such as infringement by massive corporations or internet web crawlers will become increasingly important. I think this could be achieved through education on tools like Glaze and Nightshade, which are types of “invisible filters” you can apply to your artwork for free which work to heavily disrupt or damage AI models that are trained on your work without consent.
Jia Liu of Guangzhou Huali College draws the following conclusion in a paper written for the 2025 International Conference on Generative AI and Digital Media Arts:
“Looking ahead, generative AI is expected to exert increasing influence on art education – shifting the focus from manual technique to creative direction and digital literacy – and reshape the structure of the art industry by enabling scalable content production and mass customization”.
While there are good aspects to this approach, for example, Creative Direction is a broad but desirable set of skills that will help inform your practice as an illustrator, it seems to suggest that in an effective art education manual technique is not an important aspect. I think that this is a massively damaging to an effective education in the arts, as it is in the growth of an artist's personal process and practice that they gain their unique “artists handwriting” and perspective that they can use to tackle any given brief. To remove importance from this part of education would be to massively accelerate the rate at which generative AI homogenization would nullify their unique selling point as an artist.
A different approach for the artist themselves to be better equipped to understand the value and ethics present in their own work could be better education around the philosophies related to art value. Common arguments for the use of generative AI can make the artist feel as though it may be pointless to resist – that the machine will outpace any development they could have and replace the value they could provide. Even if it may seem obvious to some, it is important to know and be educated on how the process of refining your personal practice and improving your skills as an artist can be an immensely rewarding experience, and it is within that effort that valuable art and stories are able to be created.
In a broader sense, because of the way in which automating creation can devastate the individual's capacity for meaningful expression and connection, I think we will see an aesthetic shift towards deeply textured mediums. Multi-media solutions and other methods which clearly display the amount of time and effort that went into a piece of art will become preferred by both brands and consumers who will seek out signs of authenticity in a world where any given visual art could have been created by a low-effort and illegal process. This is supported by various studies, institutions and individuals.
Above: Designs for WIM Bakery, Kevin Högger, 2025.
For example, in the itsnicethat Forward Thinking 2026 series, which gathers several trends and predictions from practicing artists about the year ahead, every single one is about either utilising chaos and randomness in layout, or the leveraging the imperfections present in certain media choices. On the rise of “intentional error” usage via scanning staff writer Ellis Tree writes:
“As current AI image generators still struggle with authentically replicating the nuances of a layered, mixed-media style, we think that an overworked, scanned or heavily textured approach to image-making will only grow in 2026”
While this is particularly impactful, it is important to keep in mind that Tree states that this is a weakness of current AI image generators – so to fully utilise the effectiveness of this method, it must be underpinned by extremely strong use of narrative and meaningful imagery. The official Stills design trend report points to maximalism as another standout feature of 2026 countertrends that are very strong. Even within extremely loud, bold or “full” imagery, as long as careful consideration is given to each piece of a composition, the feeling or message you are trying to convey will ultimately shine through – and the more related touchpoints you include, each with their own sematic, societal or cultural meaning, the less likely AI will be able to come close to even accidently reproducing a similar image.
Above: Selection from Stills Design Report 2026, p32.
In conclusion, generative AI is here to stay and is a continuing danger to society as a whole. I have provided evidence of the harmful effects of AI on the livelihoods of creatives today and how it can subsequently damage the capabilities of the creatives of tomorrow. While it can be daunting to stand up to the large capitalistic systems that are enabling these systems, I have discussed how we have an ethical duty to attempt to stand up for our creative spirit and legal rights through proper legal education and artistic principles. Ideally, through proper legislation and collaboration between creatives, lawmakers and programmers – generative AI models will not remain in their current incarnation, but instead will be ethically sourced and if used, used as a springboarding tool rather than complete replacement.