The images in this series are Digital Photomontages based on AI-generated portraits of non-existing people, 2022.
Artificial Intelligence (AI) is a leading force of our day and times. It is already altering the world and raising important questions for society, the economy, and us as human beings.
With the development of new technologies (the Internet, social networks, Big Data, the new generation of algorithms, applications for smartphones, videogames, etc.) the urge to improve the ways in which we use such technology and the need to address the ethical aspect of the coexistence of human beings and AI machines to protect our democracy and its values becomes more crucial than ever. Many organizations concerned with AI ethics and policy (public and governmental, corporate and societal) recommend government regulation to ensure transparency and, through it, human accountability.
The Algonauts is a collection of digital, AI-generated portraits of non-existing people appropriated from a popular Artificial Intelligence-powered website that was initially designed to raise awareness of AI’s ever-increasing power to present as authentic, images that are entirely artificial, and, more crucially, to inform the public of the potential damage it could cause.
In the case of The Algonauts project, a set of such portraits has been “artistically” altered – i.e., through a privacy-protection, censorship-assertive act of artistic, human-driven mania – with the addition of a series of graphic masks made out of stylized sets of children’s eyes, generated by the same algorithmic force, and specifically tailored by the author as a means of concealing the identity of all portrayed non-existing subjects.
The platform is based on a fascinating technology that allows an algorithm known as a generative adversarial network (GAN) – a type of AI developed by Nvidia in 2018 that learns to produce realistic but artificial examples of the data it is trained upon – to generate an endless series of realistic faces of non-existing people each time the page is refreshed.
In a 2021 paper titled “This Person (Probably) Exists,” researchers show that numerous faces produced by GAN bear a striking resemblance to actual people who appear in the training data. The scientific study demonstrates how reversing the algorithm can effectively unmask the real faces the GAN was based upon, making it possible to expose the identity of those individuals with inevitable nefarious consequences in terms of privacy and safety. We are at the gates of mock subjective states, and AI is leading the way. Yet AI has only arrived at those gates because human beings have decided that the virtual is more real than the Real.
The Algonauts aims to symbolically correct and mend the privacy flaw and serves as a reminder of how crucial it is to preserve and nourish “human identity” when dealing with AI.
While AI has already taken over human tasks in several fields, these are areas where the work processes merely involve the execution of commands, and in which no interpretative skills are required. Or so it is said. AI and humans do not have the same qualities and abilities. AI always follows a predetermined pattern, the previously programmed algorithm, to reach the same conclusions. AI-based machines are fast, more accurate, and consistently rational, but they are not intuitive, emotional, or culturally sensitive. The strengths of the human being lie exactly where the weaknesses of AI lie – and vice versa. It is the interaction of both that is most effective in achieving optimal success. Therefore, the key to making AI work is human insight, contextual awareness, and creativity.
The Algonauts series intends to question the controversial idea of the complete replaceability of humans by Artificial Intelligence and suggests a reflection on the necessity to embrace raising awareness, improving the culture, and creating adequate regulation, rather than passively demonizing algorithmic technology.
In the era of deep fakes, maybe these portraits of AI-generated people with counterfeit, childish eyes staring at us in a mirror-like dynamic are more accurate than many social profiles of our days.
© Gaia Light 2022