크리스티나 로페즈 (Christina Lopez), <초상화(대리인들) (Portraits (Proxies))> (2020)

단채널 영상 / 컬러 / 무음, 30초 GAN(생성적 적대 신경망)을 활용하여 필리핀 내 여론 조작 세력의 이미지를 합성, 실존하지 않는 허구의 초상들을 생성한다. 노이즈 가득한 이 초상들은 정체성이 신자유주의적 권력에 의해 무기화되는 양상을 상징한다. 기술적 접근의 한계와 생산 수단의 부재로 인해 더 이상 제작이 불가능해진 본 영상은, 데이터화된 신체와 권력 구조, 그리고 기술적 소멸의 문제를 함축적으로 보여준다.

Christina Lopez, Portraits (Proxies) (2020)

Single-channel video / Color / No Sound, 30 secs

This work uses a machine learning framework called GAN (generative adversarial network) to make portraits of people who don’t exist, composite identities generated from hundreds of profile pictures of paid trolls* operating in the Philippines. These portraits appear disembodied. They are noisy and chaotic, reflecting how identities are often co-opted and weaponised by neoliberal governments, imperialist forces, or capitalist endeavors. The work is a study on portraiture, image-making’s modes of production, and of identity and the body in relation to power.

Because the technology used is dated (pre-diffusion image generation), and I have to rely on tools like google colab to use machine learning models since I do not own the means of production, the work is now semi-obsolete. I have been locked out of the GAN, and can no longer produce more portraits. This video is perhaps the last clip I have relating to the work, until I can find a way to access it again.

*Paid trolls are proxies in the sense that they are surrogates tasked by a more powerful body or entity to spread (mis)information and interact with other people online. In the Filipino context, there are enough of them to impact news, public sentiment, elections, etc.