AI in Entertainment Media: An Ethical Analysis (2023)
Time
Spring 2023
Context
School project
Written for a bachelor's course, this ethical analysis examines a case study featuring Netflix Japan's The Dog and The Boy (2023), an AI-assisted media production, and the public debates surrounding it.
...
Ethical decision making applied
This paper critically discusses the ethical aspects related to a media case. A summary of the case will be introduced, focusing on the ethical concerns revolving the media production. Different players and values involved will be identified and respectively analysed with the Potter Box, which is a guideline of ethical thinking developed by Ralph Potter Jr. (Peck & Reel, 2013, p.35). Finally, the paper will conclude with a final moral decision suggested by the writer.
Media case and ethical dilemma
In January 2023, video production company Netflix Japan announced and released “The Dog and The Boy”, a three-minute-long mini-film in which all background scenes were created mostly by AI image-generation technology (Netflix Japan, 2023b). As shown in the film’s ending credits, the background creation process is a combination of applied AI technology and human animators who created the first draft and finalising touches (Netflix Japan, 2023a). The film is a collaborative work between Netflix Japan, Japanese animation company WIT Studio, and Japanese AI company Rinna Inc.
According to an official article by Netflix Japan (2023b), the company described the film as an “experimental” creation which seeks to explore possibilities of applying AI technology to address the shortage of animators in the domestic anime industry. In the same article, the three directors who oversaw the project all express positivism towards the film, stating that utilising the latest AI technology as a creative tool opens new possibilities to the traditional process of animation production. Moreover, they trust it to also help speed up creation time, reduce stress for human animators, and would allow animators to fully devote themselves to more sophisticated tasks such as portraying characters, expressions, and storylines—which they believe to be key creative components “only humans are capable of” (Netflix Japan, 2023b).
In response to the company’s Twitter post on the film, however, were intense online debates where netizens questioned, among many things, the morality of the producers for using AI technology to create illustrations in commercial animated productions. In line with the paper’s focus on ethical aspects behind media productions, the broader ideological question of whether AI-generated art should be perceived as Art will not be addressed in the discussion, as it falls beyond the scope of media ethics.
Opposers of the project voice their disagreement toward the producers mainly due to ethical issues related to the rewards and creative consent of (1) artists whose arts are possibly used for training AI algorithms, and (2) human animators who contributed to the project. Current AI image-generation models are trained by feeding the algorithm a large number of (un)copyrighted artworks from a vast network of sources (Škiljić, 2021), which the algorithm then selects, engages, and generates new images through a complex machinic process (Maerten & Soydaner, 2023).
Because of the possibility that some of the artworks used as training sources may come from unconsented and uncredited human artists (Škiljić, 2021), many opposers regard this project, which uses “unethical” AI technology, as equally unethical. Furthermore, although efforts of human animators were undeniably made during the film’s production process, the ending credits only show the background animators as “AI (+Human)”, without properly crediting human animators by names; however, AI developers were credited (Netflix Japan, 2023a), resulting in many criticisms on how the film glorifies AI (and its developers) yet neglects the time and efforts of human animators.
Contrariwise, supporters of the project perceive the producers as well-intended, distinguishing the film’s novelty as well as contribution to the technological and industrial fields. Some recognise the project’s experimental effort to utilise human-AI cooperation in animated productions, defending that AI technology is a new technique that requires many trial-and-errors in order to mature and be better implemented into future productions. Other opinions agree with Netflix Japan’s statement on AI technology’s potential to improve animated productions, especially in terms of generating innovative visuals beyond human imagination and making the current production process more efficient by reducing artists’ unnecessary workload. Some supporters acknowledge potential copyright issues regarding AI technology; nevertheless, they conclude that the overall advantages of using it outweigh inevitable losses.
The case is a valid example of real-life ethical dilemmas in media productions as it consists of the clash of multiple stakeholders, moral standards, and interests (Kneer, 2023). Moreover, the difficulty of attributing legal creators of AI-generated artwork, and the ethical threat behind it, has also been an important yet unresolved problem in academia (Škiljić, 2021). As there is currently no unified consensus nor well-established legal regulations detailing every aspect of the copyright or commercial use of AI-generated artworks (Mathur, 2022), it can be argued that the evaluation of whether it is morally acceptable or unacceptable to use AI technology to assist in commercial animated productions, at least as of present, is indeed a challenging decision that relies heavily on the moral judgements per individual.
“…the evaluation of whether it is morally acceptable or unacceptable to use AI technology to assist in commercial animated productions, at least as of present, is indeed a challenging decision that relies heavily on the moral judgements per individual.”
Analysis
In this section, the Potter Box (Peck & Reel, 2013, p.35) will be applied in the ethical evaluation process. The process consists of four steps: (1) facts: defining the dilemma, (2) values: identifying all stakeholders who are or might be affected, including their values and interests, (3) principles: analysing significant values through established ethical principles, and (4) loyalties: identifying the (most prioritised) loyalties of the dilemma holder (Peck & Reel, 2013, p.35). The Potter Box aims to clarify reasoning until a final moral decision on the case is drawn.
Step 1: Facts
According to an official article published by Netflix Japan (2023b), three producers affiliated with Netflix Japan and WIT Studio collectively share the responsibility as the leading directors of an animated short film, named The Dog and The Boy, where AI image-generation technology has been used to assist human animators in creating all background illustrations. The main decision-makers are the three directors, and the project is a collaboration between Netflix Japan, WIT Studio, and Rinna Inc. It is shown by the ending credits of the film that not only AI, but numerous staff (presumably across all three companies) were also involved in the actual production of the film, but none are properly credited individually by name. Instead, the ending credits only attributed “AI (+ Humans)” as the animators (Netflix Japan, 2023a).
Given the film's commercial nature, most details on the actual AI technology used, as well as the sources of images used for AI model training, are fairly vague. Based on existing literature and known facts, there are currently no legal issues with using AI image-generation technology in their field (Mathur, 2022), nor has the film legally infringed anyone’s copyright. Despite that, however, the film still sparked much controversy online as public perception of the morality of the application of AI technology in commercial animated productions differs significantly. Valid arguments are identified as follows:
1. Opposers of the project regard it as an unethical project mainly due to (1) the possible threat of ignoring artists’ consent and credits surrounding the current technology behind AI training models, and (2) the efforts of human animators who worked on the film’s background were not properly credited as they should be by right.
2. Supporters of the project acknowledge its contribution to the future development of AI technology from both technological and industrial perspectives. Overall, they do not deem the project immoral, or not enough, in comparison to its positive contribution to society.
3. The interests of numerous artists may or may not be disregarded, as it is possible that parts of the film may come from the AI algorithm which has been trained with artworks created by unconsented and uncredited artists.
Finally, copyright and consensual issues behind AI art-generation technologies are academically proven ethical hazards (Škiljić, 2021), but there is currently no universal consensus nor legal enforcement on the use of AI in commercial animated productions, opening room for moral debates.
The paper regards the directors (main decision-makers) of the film as dilemma-holders who control the situation. The dilemma is identified as whether they are ethical in their appointed implementation of AI image-generation technology in the animated film, The Dog and The Boy.
Step 2: Values
Important values of the dilemma holders and (potential) stakeholders are identified as follows:
Innovative purpose of using AI technology in animation production. The directors have the right to challenge and innovate in their commercial products. The project is also one of the first in their field to put AI in the commercial test, meaning they could likely contribute to the whole animation industry with their results.
Improving effectiveness in production. It is logical, being animation directors, to find possible means such as AI to decrease time and resources for production. This and the first value are both important to the producers and the supporters of the film.
Ensuring transparency of the creative process. The producers supervise the project, hence are partially responsible for clarifying all necessary information to the public regarding their decisions in case valid questions or concerns arise. Moreover, influential guidelines regarding the matter also suggest developers be transparent on how AI algorithms are trained, and all possible societal impacts the product may create (UNESCO, 2021). This value should be reflected by the producers, or the perceived credibility of the film, themselves, and their companies could be damaged.
Protection of artists for their work. Artists’ efforts should always be recognised, credited, or rewarded when necessary; the interests of individual artists should always be protected. This value is currently difficult to achieve as far as AI image-generation technology is concerned but remains valid to producers, artists, and the public.
The human animators’ right to be properly credited. This is similar to the former value but refers to all animators whose work and efforts contributed to the film. It is partly the directors’ job to ensure the interests of all individuals in their production team are properly returned; otherwise, it is unethical. Although the directors want to emphasise AI contribution, they should not overlook the human animators who unquestionably contributed too.
Step 3: Principles
In this stage, established ethical principles are drawn to give moral guidance to evaluate what should (or should not) be done in the case.
Mill’s Utilitarian perspective. It is mostly understood in Utilitarianism that the best action is one that allows the most pleasure (such as happiness, entertainment, and public good) for the biggest number of individuals (Rachels & Rachels, 2012a). However, according to Mill (1863), it is equally important to consider that (1) every individual’s pleasure should be equally considered; (2) there are different kinds of pleasures which can, and should, be ranked and prioritised by their importance; and lastly, (3) consequences of one’s action is the utmost important (instead of intentions). Considering this, the directors still should not act in favour of the individual pleasures of the majority (global audiences’ entertainment and industry members’ enlightenment, respectively), because such action may sacrifice the pleasures of the minority (artists’ agencies, creativities, and credits), which are of higher moral importance. Furthermore, choosing to develop an AI-assisted film with good intentions does not justify the possible consequences stemming from the uncertainties of AI technology. Hence, the Utilitarian perspective suggests the directors’ actions as unethical.
Kant’s Categorical Imperative. In essence, Kant suggests that we should never use others as means to an end, regardless of the consequences (Rachels & Rachels, 2012a). Thus, the producers should always treat every contributor equally as good. They should never permit the possible use of some artists’ work as a means to develop AI algorithms, which then contribute to the creation of an innovative film attributing only themselves. This also implies that the directors’ actions are unethical.
However, there is no actual proof that other artists’ works were actually appropriated to train the AI model used to produce the film; no information is given, nor can anyone easily track all artists and works “borrowed” by these AI algorithms.
Held’s Ethics of Care. Held’s principles reject that all individuals evaluate others by the same rational standards; instead, it highlights the significance of situational factors and individual relationships in guiding our moral compasses, and certain relationships such as care, trust, and affiliation, are vital components to evaluate when addressing moral dilemmas (Rachels & Rachels, 2012b). This perspective suggests that rather than aiming for achievements such as innovation or effectiveness or focusing on the interests of unknown artists whose works may or may not be unethically used by the AI algorithm, it is perhaps more meaningful and of utmost priority for us, as the directors of a project with responsibility to care about others, to make the best decision where individuals who require care, especially in our production team, are properly treated. In this regard, the producers fail to ensure animators are properly attributed in the ending credits.
Step 4: Loyalties
Here, the directors should decide whom or what values they are most loyal. Ethical principles show that it is possible that AI technology is problematic and should not be used in this context, so the film should not be made (at least not by AI). But because copyright and attributing matters behind AI technology are currently beyond our capability to solve, we may still succeed in using it with caution by being loyal to our professional responsibility, while paying extra care about the welfare of known human actors who are involved in the production. Loyalty to the ethical guidelines of AI development will lead us to act explicitly transparent about the production process, sharing as much information as possible to educate and assure the public, and minimise ethical trepidations related to ambiguity and bad communication.
Conclusion
“If [Netflix Japan’s The Dog and The Boy] is handled differently by responsible actors in the future, with more care, transparency, and better communication on the production process (especially on how AI technology works and its consequences), paired with a comprehensive ending credit, then their actions may be regarded as (more) ethically sound.”
The directors were perceived to be predominantly concerned with advocating the industrial values of their production, with little to no actions taken to ensure transparency, public accountability, or the welfare and interests of other stakeholders, particularly the uncredited human animators. This suggests that their supervision of the use of AI image-generation technology in the film's production is unethical. However, it is not to say that AI-implemented projects as such will always be unethical and should never be used in media productions, because ethical decisions always stem from specific contexts and interpersonal relationships involved, as suggested by Held’s Ethics of Care (Rachels & Rachels, 2012b). If this project is handled differently by responsible actors in the future, with more care, transparency, and better communication on the production process (especially on how AI technology works and its consequences), paired with a comprehensive ending credit, then their actions may be regarded as (more) ethically sound.
In the face of fast changing development of AI applications in the commercial sector, as well as the policies and social norms surrounding them, careful implementation of AI technology, combined with critical study of the complexities and ethical consequences of each player involved, appears to be a most comprehensive way to address the topic.
February 2023
...
References
Kneer, J. (2023, January 29). Foundations: Media, ethics and morality [PowerPoint slides]. Erasmus University Rotterdam.
Maerten, A. S., & Soydaner, D. (2023). From paintbrush to pixel: A review of deep neural networks in AI-generated art. https://doi.org/10.48550/arXiv.2302.10913
Mathur, A. (2022, Novermber 21). Art-istic or art-ificial? Ownership and copyright concerns in AI-generated artwork. Center for Art Law. https://itsartlaw.org/2022/11/21/artistic-or-artificial-ai/
Mill, J. S. (1863). What utilitarianism is. In Utilitarianism (pp. 4−17).
Netflix Japan. (2023a, January 31). アニメ・クリエイターズ・ベースアニメ「犬と少年」本編映像 – Netflix [Video]. YouTube.
Netflix Japan. (2023b, January 31). Netflixクリエイターズ・ベース、rinnaとWIT STUDIOとの共同制作プロジェクト、アニメ「犬と少年」を公開。クリエイター支援の可能性に一手を。 Netflix. https://about.netflix.com/ja/news/the-dog-and-the-boy
Peck, L. A., & Reel, G. S. (2013). The Potter Box. In Peck, L. A., & Reel, G. S. (Eds.), Media ethics at work: true stories from young professionals (pp. 35−36).
Rachels, J. & Rachels, S. (2012a). The elements of moral philosophy (7th. ed.). Second Story Press.
Rachels, J. & Rachels, S. (2012b). Caring relations and principles of justice: Virginia Held. In The right thing to do: Basic readings in moral philosophy (pp. 74−80). McGraw-Hill.
Škiljić, A. (2021). When art meets technology or vice versa: Key challenges at the crossroads of AI-generated artworks and copyright law. International Review of Intellectual Property and Competition Law, 52, 1338–1369. https://doi.org/10.1007/s40319-021-01119-w
UNESCO. (2021). Recommendation on the ethics of artificial intelligence.https://unesdoc.unesco.org/ark:/48223/pf0000380455
< back to writings / research