By Marina Hassapopoulou
A very memorable visit to the traveling exhibition Synthetic Memories /Memorias Sintéticas at the Design Museum of Barcelona, part of the Synthetic Memories initiative by Domestic Data Streamers that is aimed at recovering undocumented or lost visual memories using AI image generation.
The Public Office of Synthetic Memories is an installation that invites members of the public to take part in recovering and reconstructing their memories. The exhibition, Synthetic Memories, is an immersive display that merges technology and personal stories, featuring the images created in this office thanks to artificial intelligence (AI) prompted by city residents’ memories. And the repository will keep growing as residents who are interested participate in the project… –> download PDF with more info about the project and visit the Synthetic Memories website.
Scroll down for Marina’s reflections…
What is a Memory if not Synthetic? Some scattered reflections via a trajectory of my own work in technomemory studies:
At their core, many of the technophobic apocalyptic fears about AI have to do with its [arguable] potential to replace humans, including human memory and even the capacity to think. Striking a productive balance between seeing the productive cultural contributions of AI, and a healthy critical stance towards AI’s own non-human autonomy, the Synthetic Memories project is a fascinating initiative that indicates ways of co-producing with generative systems. The exhibition that showcases some of the outcomes of the project is set up like an old-school analog archive (the empty file containers are nothing but props, as I discovered) with an old computer station that informs the visitors about the objectives of the project and its functionality.
Refreshingly, Synthethic Memories does not consider AI as the ultimate epistemological paradigm shift. Instead, as the informational video conveys, AI is nothing but a tool in the scheme of a larger nuanced, human-centric memory project. The video directly undermines generative AI’s own inherent intelligence (at least as we humanly define it) by calling it a “statistical machine on steroids” that generates “vectors of memory.” With this in mind, AI is regarded as a tool that could serve as a stand-in, rather than a replacement, of many types of mnemonic conditions including memory loss, trauma, preservation, and wishfulness. I would call it a “mnemonic prosthesis” that, in some ways, does not differ from photographs, illustrations, films, social media, and other memory aids in our contemporary memex landscape. Synthetic Memories does not necessarily aim to fill certain gaps of human memory; rather, it non-conclusively points to those gaps and critically grapples with them.
A mix between creative expression, cultural memory preservation, and scientific research, the Synthetic Memories project began in 2022 with pilot experiments in nursing homes that are using reminiscence therapy for dementia patients, then expanding to the preservation of memories from migrant and refugee communities who had lost their photographic mementos, and then traveling to neighborhoods such as the Bom Retiro (in São Paolo) to visually reconstruct the memories of its multicultural immigrant population. In some aspects, the project reminded me of other initiatives such as the Holocaust survivors project that used generativeAI to revive and revision survivors’ oral testimonies.
In my new book, I reflected on cultural co-creation as a productive means of collaborating with “smart” machines, with some ambivalent conclusions (if we can conclude at this point) that may not fully apply to Synthetic Memories but are certainly worth considering when it comes to this type of memory work:
“On the one hand, this use of AI has prompted me to reflect on the productive and poignant aspects of co-creating with technology to produce new historiographical tools and invaluable contributions to collective memory and memorialization. On the other hand, such uses of AI–where the AI program […] was used to convert the Holocaust survivors’ oral testimonies and personal photos into text prompts and then into graphic pictures–could mean that that AI aesthetics could become the new cultural vernacular and the new form of hybrid technomemory that includes standardized interfaces and a machine-assembled cultural database. The use of AI to preserve and visualize memory overlaps with other methods such as the use of film and special effects to supplement historical deficit when it comes to suppressed histories that lack visual evidence. Conversely, though, the use of AI for visual testimonies and ocularcentric memorialization could further exacerbate the dominance of the visual as a truth-index and its primacy in the historical archive. Disturbingly, I noticed that AI tools can now agnostically apply a cinematic sepia-toned ‘holocaust aesthetic’ to other disasters when overlapping prompts are used, which could lead to the homogenization of collective memory and the aestheticizing of trauma…” (208-9).
Some years ago, to dispel the hype surrounding VR as the ultimate “empathy machine,” I revisited Alison Landsberg’s work on prosthetic memory — a concept and cultural process that is certainly even more relevant to a project like Synthetic Memories and to generativeAI — in relation to docugames (documentary games) where “the historical is perceived in a composite manner” rather than in a photorealistically or factually driven mode. One of my objectives in that article was to examine any emerging technology on a “continuum with the non apparatus-specific cultural traditions of older media […to explore the] profound implications that historical changes in the apparatuses of vision have had on cultural perception and interpersonal communication. Analyzing certain strategic continuities in the cultural status of these media is a more productive approach than placing [them] on a deterministic timeline of technological evolution…” Synthetic Memories’ composite aesthetics– a mixture of family-photo album-style and surreal aesthetics (with a signature Dali quote to endorse it!)– makes us see the past, and more specifically memory, in a layered way rather than as seamlessly assimilable into the personal and collective cultural fabric.
In my docugames article, I wrote: “Landsberg uses the term ‘prosthetic memory’ to refer to the process by which individuals relate mediated images of the past to their own life experiences to the point where those images become experientially internalized as personal memories. Prosthetic memory, then, overlaps with the notion of technomemory: the fusion of technological tools with the biological functions of information accumulation, recollection, and sociocultural associations to create hybrid memories.[i] Landsberg adds that, ‘in its most progressive versions, prosthetic memory creates a feeling for, while feeling different from, the other, thereby permitting ethical thinking.’ The ‘strategic remembering’ of national trauma and collective history afforded by mass media platforms can, according to Landsberg, induce a sense of ‘collective social responsibility,’ which she recognizes a productive way of producing ‘potentially counter hegemonic public spheres.’ “[ii] Fast forward to 2024, the counter hegemonic has become integrated into the hegemonic as new cultural memory tools converge with corporate platforms and participatory strategies…
This idea of “counter hegemonic public spheres” was particularly evocative to me when I experienced the Synthetic Memories project in Spain, a country that has a historically traumatic relationship to collective memory and the act of remembering/forgetting. The historical hauntings of Spain’s 1975 Pact of Forgetting (Pacto del Olvido), the political decision by both left and right-wing parties to avoid remembering the atrocities of the Franco regime in order to more easily transition to a post-totalitarian era, still ghost contemporary memory projects, at least to a cultural outside like myself with her own country’s (Cyprus) traumatic history [unbiased link unavailable]. It wasn’t until 2007 that Spain’s Historical Memory Law was passed, officially recognizing the victims on both sides of the Spanish Civil War (1936-39) and Franco’s dictatorship, and acknowledging the civic rights and moral reparation of those who were persecuted for their political, ideological, or religious beliefs. Interestingly, the use of GenAI has made room for the excavation of collective history through the recollection of personal memories; for instance, in a recording of memory #38 (see photo), Rosa remembers the day she and her fellow students threw Franco’s statue down the University of Barcelona stairs over 70 years ago. To me, it is still strange to see Franco’s name mentioned in any type of exhibition in Spain (like Hitler in Germany), so this exhibition revealed a potentially less traumatic or at least non-politicized way to revisit collective history through the GenAI exploration of personal memory; in this way, the historical initially appears as a backdrop to individual memories, but to the casual observer takes on a life of its own due to its beyond-individual resonance.
Overall, I appreciated Synthetic Memories‘ critical approach to generativeAI and its mindfulness of ethics (beyond just the now-token gesture of mentioning “ethics” abstractly). The project embraces what it calls the “digital nostalgia” aesthetic (courtesy of DALL-E 2 and other GenAI tools) that is deliberately blurry, and also encourages memory participants to critically engage and edit the images once they are generated. Lastly, I appreciated the project’s nod to smell, even if it was just mentioning that these “vectors of memory” are (currently?) limited to particular senses that emphasize visuality above everything else. I am wondering how this over reliance on the visual impact memory studies, even as we are now increasingly using technologies that cannot “see”…
*Will return to this post for a more cohesive essay soon…*
What do you think about GenAI’s role in cultural and personal memory? What are some of the most successful case studies/uses you have seen so far? The sci-fi imaginary has once again predicted this notion of technomemory — see, for example, the TruNode brain-computer interface in Alex Rivera’s Sleep Dealer (2008), among many other examples…
Comment below:
[i] Marina Hassapopoulou, “Interactive Pedagogy: Social Media Repurposing in Cinema Studies,” The Journal of Digital Media Arts and Practice (2016).
[ii] Alison Landsberg, Prosthetic Memory: The Transformation of American Remembrance in the Age of Mass Culture, 150-52
Expressive AI 
Leave a comment