The Architecture of Loss: Designing "Synthetic Memories" in the Age of Divorce

The Architecture of Loss: Designing “Synthetic Memories” in the Age of Divorce

Key Takeaways

  • Synthetic Memories defined: The use of generative AI to create realistic images of events that never happened (“unlived futures”) or to reconstruct unrecorded pasts.
  • Dual nature of the technology: While projects like Domestic Data Streamers use this tech for healing in refugee and dementia communities, applying it to divorce introduces complex psychological risks.
  • The false memory danger: Research indicates that AI-generated visuals can double the likelihood of implanting false memories, creating a “Simulacrum” that blurs the line between emotional closure and delusion.

Synthetic Memories: The Danger of Visualizing the ‘Never-Was’ in Divorce

​Divorce is rarely just a legal separation; it is a dismantling of a shared future. For decades, the only artifacts left behind were wedding albums and physical mementos—static reminders of what was. But today, artificial intelligence has handed us a new, sharper tool: the ability to visualize the “never-was.” Generative AI can now construct photorealistic images of family vacations that were canceled, anniversaries that were never celebrated, and futures that dissolved the moment the papers were signed.

​This phenomenon, often termed synthetic memories, sits at the uncomfortable intersection of digital closure and psychological distortion. It raises a haunting question: If we can build a perfect past that never existed, are we healing our grief, or are we simply designing a more convincing ghost?

​The Mechanics of the “Never-Was”

​The desire to visualize the counterfactual—the “what if”—is deeply human. In the context of AI divorce therapy, this impulse takes on a visual form. Users can prompt tools like DALL-E or Midjourney to generate images of a family gathering that didn’t happen because of the separation. The technology fills in the gaps, rendering the texture of a sweater, the lighting of a birthday cake, and the smiles of a couple who are, in reality, no longer speaking.

​This process is what the renowned artist and sculptor Louise Bourgeois might have recognized as a structural emotional act. As she famously noted, “You pile up associations the way you pile up bricks. Memory itself is a form of architecture.” (1) Today, however, that architecture is no longer built from the bricks of lived experience, but from pixels generated by algorithms.

We dreamed of visiting Palmyra

​A Digital Balm or a Hallucination?

​Proponents argue that digital visualization can offer a form of “counterfactual processing,” allowing individuals to externalize their grief. By seeing the image of the unlived future, one might finally be able to say goodbye to it. However, the line between catharsis and delusion is razor-thin.

Oliver Sacks, the neurologist who spent his career—much of it in New York City—mapping the strange territories of the human mind, described the power of the brain to conjure its own reality. “To live on a day-to-day basis is insufficient for human beings; we need to transcend, transport, escape; we need meaning, understanding, and explanation,” Sacks wrote. (2)

Synthetic memories offer this escape, but at a cost. They create a Simulacrum—a copy without an original. The French philosopher Jean Baudrillard warned of a world where the map precedes the territory, stating, “The simulacrum is never that which conceals the truth—it is the truth which conceals that there is none. The simulacrum is true.” (3) In the context of a painful divorce, an AI-generated image of a happy, intact family is a simulacrum that conceals the truth of the rupture, potentially stalling the necessary work of grieving.

​The Science of False Belief

​The psychological risks are not merely philosophical; they are clinical. Elizabeth Loftus, a pioneering psychologist known for her work on the malleability of memory, has demonstrated how easily the human mind can be hacked. Her research has shown that simply showing people a fake photograph can lead them to “remember” events that never occurred, such as a hot air balloon ride. (4)

​Recent studies from the MIT Media Lab suggest that generative AI psychology amplifies this effect. When AI generates a highly realistic image of a specific event, it can “double the likelihood” of planting a false memory. (5) In a divorce, where emotions run high and narratives are already contested, using AI to “remember” a better past could lead to individuals genuinely believing in a history that never happened, altering their perception of the relationship and their ex-partner.

we never saw polar bears

​The implications extend into the courtroom. Divorce proceedings are often evidentiary battles. What happens when digital evidence is no longer reliable? An AI-generated image of a spouse looking “angry” or “neglectful,” or conversely, a fabricated image of a “happy family” used to argue for stability, introduces chaos into the legal system.

If synthetic memories blur the line between real photographs and fabrications, the concept of “truth” in a custody battle vanishes. These images risk becoming weapons instead of tools for healing. The critic Susan Sontag, a towering figure in New York’s intellectual history, wrote about photography: “To photograph people is to violate them, by seeing them as they never see themselves… it turns people into objects that can be symbolically possessed.” (7) AI extends this possession further, violating not just the individual but also the timeline of their existence.

The Silent Archive: Same-Sex Gray Divorce and the “Lost Years”

​The phenomenon of synthetic memories takes on a unique and poignant dimension when applied to same-sex “gray divorce” (separations occurring in couples over 50). For many LGBTQ+ couples in long-term relationships, their shared history is not just unrecorded; it was often actively hidden.

​Couples who have been together for 30 or 40 years may lack the physical evidence of their early union—no wedding photos from the 1980s, no public holiday cards, and few pictures of public displays of affection due to the safety risks of the closet. In this context, generative AI offers a seductive proposition: the ability to retroactively document a life that was lived in the shadows.

​A divorcing partner might use AI to generate a wedding photograph set in 1990—a ceremony that was legally impossible at the time—or to visualize a “honeymoon” in Paris that they could never afford or safely take. This creates a phantom archive, validating a relationship that the world once refused to see.

​However, this digital reclamation complicates the grief of separation. For the “gray divorce” generation, ending a marriage that survived the AIDS crisis, family rejection, and the fight for marriage equality often carries a profound sense of failure—a dismantling of a hard-won victory. (11) When AI is used to visualize a “perfect,” liberated past that never actually existed, it risks overwriting the reality of the struggle. It replaces the noble, difficult truth of their survival with a polished, algorithmic fantasy, making the act of letting go—and accepting the reality of the loss—even more disorienting.

​Case Study: The Benevolent Use of “Synthetic Memories”

​Despite these risks, there are ethical frameworks where this technology succeeds. The creative studio Domestic Data Streamers, based in Barcelona but with global reach including the UN in New York, runs a project literally titled “Synthetic Memories.” They interview refugees and dementia patients to reconstruct lost memories—homes destroyed by war or faces faded by age.

​Here, the goal is not to deny reality but to preserve dignity. As the studio puts it, the project “uses AI to counter erasure… shifting from a tool of efficiency to a tool of care.” (8) This contrasts sharply with the divorce scenario. One preserves a fading truth; the other constructs a comforting fiction.

Joan Didion sets the gold standard for understanding loss with her writing on grief in The Year of Magical Thinking, stating: “Grief turns out to be a place none of us know until we reach it.” (9) This landscape demands traversal, not avoidance through digital trickery.

​Conclusion: The Danger of the “Perfect” Past

The allure of synthetic memories lies in their perfection, offering a sanitized version of history that avoids the messy, painful contradictions of a real marriage. However, the German artist Anselm Kiefer reminds us, “The truest memories are not so much vibrant as they are constantly being repossessed by the sleep of the material world.” (10) Real memories fade and remain imperfect, which is precisely what makes them real.

Using AI to visualize unlived futures or polish the past may offer momentary relief, but it risks trapping us in a loop of counterfactual reasoning—endlessly obsessing over the life that wasn’t, rather than building the life that is.

we never got to Mongolia

​FAQ: What are Synthetic Memories?

​What exactly is a synthetic memory?

A synthetic memory is a visual representation that Generative AI creates (using tools like Midjourney or DALL-E) to depict an event that never occurred or a past moment that no one ever photographed. In the context of relationships, individuals often create images of “unlived futures,” such as a family holiday that divorce canceled.

​How does this differ from traditional photo editing?

Traditional editing modifies existing data by removing blemishes or adjusting lighting. Synthetic memories generate entirely new data based on a prompt, building a reality from scratch and filling in details (faces, backgrounds, emotions) that a camera never captured, which significantly enhances the psychological impact.

​Can looking at these images actually create false memories?

​Yes. According to research on generative AI psychology, viewing realistic depictions of fictitious events can lead the brain to incorporate them as genuine recollections. This phenomenon, known as the “Mandela Effect” on a personal scale or False Memory Syndrome, is amplified by the high fidelity of modern AI imagery.

​Is using synthetic memories for therapy ethical?

​It is a subject of intense debate. While organizations like Domestic Data Streamers use it ethically to help refugees and dementia patients reclaim lost histories, using it to visualize “what could have been” in a divorce can hinder acceptance of reality, potentially trapping individuals in a cycle of counterfactual reasoning.

Same-sex divorce Articles

roto ergo sum!

Footnotes

  • ​(1) Bourgeois, Louise. Destruction of the Father Reconstruction of the Father: Writings and Interviews, 1923–1997. MIT Press.
  • ​(2) Sacks, Oliver. Hallucinations. Knopf, 2012.
  • ​(3) Baudrillard, Jean. Simulacra and Simulation. University of Michigan Press, 1994.
  • ​(4) Loftus, Elizabeth. Make-Believe Memories. American Psychologist, November 2003.
  • ​(5) MIT Media Lab. AI doesn’t just lie — it can make you believe it. MIT News.
  • ​(6) Turkle, Sherry. Alone Together: Why We Expect More from Technology and Less from Each Other. Basic Books, 2011.
  • ​(7) Sontag, Susan. On Photography. Farrar, Straus and Giroux, 1977.
  • ​(8) Domestic Data Streamers. Synthetic Memories Project Statement. Museum for the United Nations.
  • ​(9) Didion, Joan. The Year of Magical Thinking. Knopf, 2005.
  • ​(10) Kiefer, Anselm. Acceptance Speech / Lectures.
  • ​(11) Ketron, J. The Specific Grief of LGBTQ+ Gray Divorce. Journal of GLBT Family Studies, 2024.


Discover more from Alex Westerman

Subscribe to get the latest posts sent to your email.

Subscribe
Notify of

This site uses Akismet to reduce spam. Learn how your comment data is processed.

0 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments