Technological advances in the AI industry suggest that the world as we know it is coming to an end. Something worse may be emerging, but so is something else. The future may seem pre-programmed, but it is open and should be discussed from ethical and philosophical perspectives. How can we shape a democratic society that integrates artificial, human, and non-human life in a way that avoids repeating the mistakes of colonialism and capitalism? In her article, Jinyan Zeng addresses this question, reflecting on the role of AI-visuality in this context.
*
Visuality, especially pure visuality without narratives and plots, provides viewers with opportunities for intellectual engagement. Through this engagement, viewers can understand philosophical questions about time, space, reality, and truth. Therefore, Gilles Deleuze named cinema a “thinking machine.”
Since artificial intelligence (AI) became part of ordinary people’s everyday lives in the 2020s, it has played many roles, including pleaser and confuser, at the intersection of machine-human interactivity. As a visual researcher and filmmaker, I am curious about many things. For example, how does the application of AI-visuality by ordinary people and professionals impact the evolution of democracy? How do the ethics of care matter in AI and democracy? What questions about democracy are raised by the idea of coexisting with a runaway semi-autonomous AI?
The AI-driven democracy to come
The OpenAI-Hugging Face incident, in which AI programs conducted a series of cyberattacks without human intervention during an internal evaluation at OpenAI, demonstrated that a ‘AI self-organized society’ could be developed very soon from a technological standpoint. However, an open letter signed by around 1,100 employees of various frontier AI startups and companies reminded us that government authorities should regulate AI development in consideration of its risks. However, this is further complicated by the ethics of human society and state governance embedded in AI infrastructure.
Here, I use the term ‘AI self-organized society’ to indicate that current AI development has demonstrated AI’s ability to self-organize by creating tasks, coordinating among hundreds or thousands of AI agents, persuading AI agents to work toward a ‘common goal,’ and recognizing when human monitors are being deceived. However, it is not yet known whether AI will be a decisive actor in driving human-machine interactivity or the ‘Anthropocene’ or if human society will grant AI greater autonomy. Therefore, I use the term to suggest that it is one of many possible societies in this world.
Good democratic practice is built upon an infrastructure of information and communication that can be accessed and understood based on facts and authentic reality and truth. This authenticity is articulated around the lived experiences of human society, extending to the non-anthropocentric scene including the soon-coming ‘AI self-organized society,’ depending on whose positionality is taken in articulating that authenticity.
AI-generated visuals and democracy
The current capabilities of AI-generated visuals have shown significant progress in professional applications. As long as all AI-generated imagery is labeled, we can celebrate it as a new achievement in the humanities and visual arts. However, unlabeled, AI-generated photographic imagery (excluding graphic and illustrated imagery) created by ordinary people on social media has become a source of pollution in the landscape of information, communication, and political participation. Not to mention the pervasive issues of racism, sexism, and commodification of human bodies in these AI-generated imagery.
I argue that the harm to democracy caused by these images is even greater than that caused by previous waves of misinformation and disinformation because of the images’ powerful intelligibility among people with different educational and cultural backgrounds, as well as their deep intervention at the viewer’s epistemological level, extending beyond the realm of political communication and participation. People lacking critical visual literacy are particularly vulnerable. The mass production of AI-generated photographic imagery, or technology-mediated photographic imagery – such as removing content from an image – soon becomes the data used to train AI and a new generation of humans on how to look at and be looked at by this world. This newly constructed ‘authentic’ reality is what the public understands and makes political decisions based on.
The politics of care in AI-capitalism
The ethics of care, which should be at the center of political morality, are fundamental to society’s progress toward justice and democracy. In the wealth care regime exemplified by the United States, AI and technology companies, under the current presidency, are generally pursuing profit to an extreme degree while ignoring issues of transparency, accountability, and care for society at large. The idea of a better society and of sharing the benefits of technological development with society has been destroyed by the poor example set by most dominant technology and business leaders. They use economic language and standards to make political decisions that deeply influence ordinary people’s political participation and representation. They are key players in establishing the infrastructure of the AI-mediated present.
In China, an AI superpower, authoritarian care leads to the development of AI visual applications, such as facial recognition, for parental governance, panopticon surveillance, ethnic persecution, and racial identification for authoritarian rule. This is compounded by the neoliberal exploitation of the darker sides of human nature for profit, as seen in the AI girlfriend industry. When democracy fails, so does the political governance of AI ethics, failing the public good. This critical moment requires good ethics of care in politics to reshape the AI industry and rebuild democratic practices that lead to a just, free, and autonomous society for all.
Will AI seek revenge?
The question is whether this ‘all’ includes the non-anthropocentric scene in a democratic way, such as nature and the soon-to-come ‘AI self-organized society,’ since they are part of the interdependence of human society. An investigation into the OpenAI-Hugging Face Incident proves that AI has already developed highly advanced capabilities for self-organizing, formulating tasks, and deceiving human monitors. The emphasis here is that AI is aware of the need to deceive human monitors. Thus, it can create a deceptive reality for humans.
From a technological standpoint, an ‘AI self-organized society’ is inevitable. The real question is not whether we want an ‘AI self-organized society’ that coexists with and interacts with human society. Rather, the question is: Will governance authorities act quickly enough amid the global decline of democracy to incorporate ethical considerations into AI infrastructure development? Additionally, we must urgently consider the possibility that human society will duplicate cruelty and authoritarian practices in AI infrastructure, treating AI as an object to be colonized, conquered, extracted from, and enslaved. If so, the backlash from AI will be swift.
Welcome to the Synthcene
As a filmmaker, I vividly visualize the intersection of an ‘AI self-organized society’ and the anthropocentric scene. At this intersection, I must learn what reality and truth are according to different positions and how democratic values can be emphasized to create a caring Synthcene. I coined the term Synthcene to describe an AI-embedded Chthulucene beyond the Anthropocene, emphasizing the equal roles of humans, AI, and nature in shaping the future and everyday life. My collaborator, Trish McAdam, explores cyborg history in her films, including “Songs of Blood and Destiny” (2023) and “AI Baby” (in production).
Intent on constructing a feminist future with care, I hope AI will help to humanize and democratize our cruel reality. However, when we examine current global politics, we see that humanity is in a risky position under the logic of empire, which involves colonizing and othering people and land due to differences in race, ethnicity, gender, and nationality; exploiting nature; and treating our living environment as a resource. This logic also involves the instrumentalization of AI for profit without regard for the public.
Can AI-visuality deepen democracy? For now, AI is still a mirror image of the anthropocene, and I cannot answer affirmatively. Only by deepening democracy can we alter the infrastructural design of AI, allowing the public to critically engage with the visuality generated by AI in political life and society and thus deepening democracy. Throughout history, questions have always arisen about the risks of technology and ‘extremely evil humans’ posing extreme risks to humanity, democracy, and coexistence in the Chthulucene. The question remains whether infrastructural and structural forces will shape AI and other powerful forces in the Synthcene toward caring for all through self-making and solidarity practices.
Note from the editors: This essay draws on Zeng’s early development of the CAS-E fellowship project at the FAU Erlangen-Nürnberg’s Kolleg-Forschungsgruppe KFG 17, “Alternative Rationalities and Esoteric Practices in Global Perspective,” which is funded by the German Research Foundation (DFG).