Since last 12 months I actually have been working on a project with my Thai mother, who emigrated to Australia with me in 1974 while she was pregnant. There aren’t any photos of this significant event. As a reparative act to fill the gaps in our family archive, I used a generative AI platform to create photos of my mother's migration story, a project I called Counterfactual Departures.
To create photo magic in MidJourney, I gathered information to create a series of prompts. I noted details of my mother's departure from Thailand – her age, whereabouts, what she was wearing, and the trimester of her pregnancy. I also included my very own interpretation of how she would have felt traveling alone to white Australia as a 30-year-old woman within the Nineteen Seventies. She barely spoke English and had no family or friends to live with my father, whom she barely knew.
My attempts to piece together the past using datasets from the Internet's collective archive have led me to query what role photography and memory might play on this latest era of generative AI.
Sarah Oscar
MidJourney and the “common image”
MidJourney, like other AI imaging platforms, creates images based on data aggregates mined from hundreds of thousands of online source photos.
Although not necessarily photographs, AI-generated images read like photographs, seem like photographs, are in comparison with photographs, and challenge our philosophical understanding of photographs.
As relatives of photography, they’re related to qualities akin to the past and memory, and even, as photo theorist Joanna Zylinska suggests, as perceptual devices that may achieve this Show us the longer term.
The problem that arises when generating images from photographic datasets is the way in which information is aggregated to form a picture as a statistical average. As artist Hito Steyerl has observed:
They represent the norm by signaling the mean. They replace similarities with similarities (…) in style and content they’re: common images.
By sampling from a spread of data sources and creating compilations, MidJourney highlights probably the most salient features of an information set, that mean value.
My mother's first photos illustrated this tendency. In the image, my mother and I are standing (in utero) in an oversized gray suit, searching of the image at a parking zone between suitcases and a wasteland of plastic bags. We appear to be somewhere in Los Angeles, portrayed within the sort of staged photography. They are generated from the statistical average of knowledge sets and may be described adjectivally as .
Sarah Oscar
Looking back at these early photos, a part of the project's problem is that I emphasized an excessive amount of what the past must have looked like. In order to alter the way in which this event was presented as a statistical technique of racial stereotyping, I had to put less emphasis on the image generated and switch image generation right into a generative strategy of dialogue and discussion.
Memory work as a generative method
The conversations we’ve got a couple of photograph, whether real or not, can represent a type of Memory work. Memory work refers to the way in which we process memories through the social relationships we construct around artifacts akin to photos, texts, objects, or family albums. It is a invaluable method for artists coping with events outside of history since the strategy of creating art serves a reparative function.
For cultural historian Annette Kuhn, this maintains dialogue in regards to the past. she writesMemory work implies that we will take a look at photos as:
Material for interpretation that should be questioned and dismantled for its meaning and possibilities.
By shifting my focus from the generative image to the dialogue it creates, I started memory work. Images were used to stimulate conversations about this significant event. This created space to acknowledge a spot between my imaginative recreation of events, my mother's memories, and the mean image created by the AI.
Sarah Oscar
This collaborative process between my mother, AI and I used to be enlightening. It enabled me to playfully close the gaps in our family biography. It gave my mother the chance to say, “No, it wasn’t like that, I got here to Australia with dignity,” and to acknowledge racist stereotypes of representation.
AI has the potential to create representative representations of ourselves and initiate dialogue about essential events and life experiences.
Sarah Oscar
From moral panic to collaboration
The moral panic surrounding generated images has been fueled within the media by the lack to differentiate between real and pretend.
At the guts of this debate is mimicry, whether it contributes to false memories, fake news, copyright infringement, or the substitute of humans with machine labor.
But what if generative AI may very well be used as a memory work tool to facilitate dialogue and interaction? This approach reconnects AI-generated images with previous social and cultural practices of photography. It offers a more intimate example of what AI can do.
