Artificial Intelligence and the Future of Palestinian Collective Memory

Palestinians commemorated the 75th anniversary of the Nakba this year at a time when the world is witnessing significant developments in communication and technology, particularly through the emergence of artificial intelligence applications. While the major companies developing artificial intelligence programs have voluntarily paused their activities for the time being, memory studies scholars have been preoccupied with the question of whether artificial intelligence will represent the future of collective memory. This article examines the potential and possibilities that artificial intelligence may bring to Palestinian collective memory in particular, by considering four programs and projects that have employed artificial intelligence to document collective memory. These are: the 535,000 Faces project,[1] an individual initiative; the From Numbers to Names website;[2] and the New Dimensions in Testimony program,[3] both dedicated to remembering the victims of the Holocaust;[4] as well as Google Arts & Culture.[5] These examples will allow us to examine the plans developed by profit-driven technology giants to manage collective memory through artificial intelligence, potentially replacing museums and archival institutions.

Over the past twenty years, the internet and digital platforms have, for the first time, enabled Palestinians to do what had previously been prohibited: publicly and collectively remember.[6] This act of remembrance has been shaped by the dynamics governing the relationship between memory and digital platforms. In particular, the Palestinians’ migration to digital platforms was driven by necessity rather than choice. To this day, as Al-Zubaidi noted in his study on digital nomads, “there is no place in ‘Israel’ or in the Arab countries where Palestinians have been permitted to establish museums, monuments, or public displays that preserve the collective features of the Nakba.”

It seems, however, that it may not be long before we see Palestinians building such monuments or museums through artificial intelligence. Although there are currently no AI programs specifically designed to engage with the collective memory of the Nakba, there is nothing to prevent AI programs dedicated to collective memory in the future from being developed exclusively for particular groups and not others. We will therefore focus on some of the programs that currently exist in order to explore what the age of artificial intelligence might open up for Palestinian collective memory.

Undoubtedly, one of the most significant commemorative practices—or mechanisms of memorialization—unavailable to Palestinian memory, including even digital actors, is the ability to remember all the victims of the Nakba and the mass massacres perpetrated by Zionist militias in Palestinian villages and cities in 1948. Palestinians have even been deprived of cemeteries or memorial markers where the victims of the Nakba can be commemorated. For example, the Israeli colonial authorities built a parking lot over the graves of the victims of the Tantura massacre,[7] who had been buried in mass graves.[8] Because mourning, commemorating, or visiting graves and memorial markers is currently not available to Palestinians, this brings us to the “535,000 Faces” project, launched by Matt Korostoff in the aftermath of the wave of COVID-19 that swept across the world in 2020. Using artificial intelligence, the project enabled the reimagining of the faces of more than half a million Americans who died during the first year of the COVID-19 pandemic. Because the number of victims was so immense as to defy comprehension, and because many people were unable to say goodbye to their loved ones, see them for the last time, or even honor them with a proper burial, the website created a monument to all of these victims, without exception. These are not actual photographs of the deceased, however, but representational images generated by artificial intelligence using the available data about the dead (age, race, gender, and so forth).[9] The website thus became a memorial within a public space, providing those affected by the tragedy with an opportunity to commemorate their dead. Visitors can view images of more than half a million faces. The creator of the website explained that these faces were intended to address questions such as: How do we mourn the deaths of people whose faces we never saw and whose names we never knew? How can we comprehend the scale of what happened? I believe that these questions require very little effort to transform them into distinctly Palestinian questions. Could Palestinians similarly revive the memory of the victims of the Nakba and see their faces—even those for whom no family photographs exist, but about whom some demographic information, such as age, gender, and race, is known? Available information could be transformed into memorials and monuments through artificial intelligence, enabling Palestinians to do far more than what is encompassed by public collective remembrance. The significance of this project lies not only in the capabilities of artificial intelligence, but also in the capacities of individuals. Korostoff created and designed the website himself, without receiving financial funding or technical collaboration from any institution. This suggests that such programs need not remain the preserve of institutions; rather, individuals—and, consequently, the public—could play a significant role in shaping and using them.

First: The Death of the Last Eyewitnesses

Reimagining and representing the victims of the Nakba is therefore one of the possibilities that artificial intelligence may offer Palestinian collective memory. It could enable Palestinians to confront one of their greatest concerns: the death of the first generation of the Nakba. To this day, Palestinian oral history of what happened during the Nakba remains the most important source of collective Palestinian memory. Palestinians have devoted considerable effort to archiving the testimonies of eyewitnesses from the first generation of the Nakba and transmitting them to successive generations to ensure the continued transmission of the Palestinian narrative and prevent its disappearance, while also challenging the Israeli assertion that “the old will die and the young will forget.” But what if the old—or the first generation of the Nakba—did not die? More than that, what if it were possible to bring the Nakba’s eyewitnesses back to life through artificial intelligence? In that case, the impending loss of the last surviving members of the first generation of the Nakba would no longer be a source of concern.

Once again, as of this writing, there are no artificial intelligence programs capable of reviving Palestinian memory. At the same time, however, Holocaust memory studies have made considerable progress in this field. Most studies point to technological, scholarly, and historical activity undertaken by individuals and institutions alike, involving those interested in the history and memory of the Holocaust.

Second: From Numbers to Names

Artificial intelligence can now be used to provide institutions and museums with previously unknown information and to update existing records. For example, software engineer Daniel Patt developed a program to identify previously unidentified Holocaust victims using artificial intelligence, which he launched under the name From Numbers to Names in April 2021. The program is freely available to the public. Users can upload a photograph of a family member, such as a grandfather or grandmother, after which the website searches for the person’s face among approximately half a million photographs. Having searched millions of faces in a matter of seconds, the site provides the user with the ten images it determines to be the closest matches to the face in the uploaded photograph. The site has, in fact, succeeded in identifying one “Holocaust survivor” by locating a photograph of her as a child. She, in turn, recognized the other people in the childhood photograph. This provided the website’s creator and the United States Holocaust Memorial Museum[10] with new information on which to base further research. Moreover, as demonstrated by the case of the woman mentioned above, the face-search technology is not impeded by changes in appearance resulting from aging.

Third: New Dimensions in Testimony

The New Dimensions in Testimony program, which uses artificial intelligence to create “virtual witnesses” who convey memories of the Holocaust, provides a useful model for understanding the possibilities and capabilities of eyewitness testimony in the age of artificial intelligence. The Shoah Foundation launched the New Dimensions in Testimony program with funding from Jewish individuals and institutions. According to the program’s website, it has been adopted by twelve Holocaust museums around the world. The program allows students and museum visitors to interact with a hologram of a Holocaust survivor and ask questions such as, “How old were you when the war began?” or “What song do you remember?” This virtual witness can respond to children’s and students’ inquiries and questions. The witness is not physically present in the museum; rather, testimonies recorded in advance on video are processed through artificial intelligence to create an interactive testimony delivered by an “eternal” eyewitness who cannot die. The capabilities of the eyewitness are not limited to being eternal and omniscient. A virtual witness can also possess the memories of more than one person and combine the memories of numerous eyewitnesses within a single testimony. Researchers Guy Pessach and Michal Shur-Ofry imagined what they called the “ultimate eyewitness” in their 2020 study. Such a witness could, for example, provide a comprehensive testimony about the Vietnam War. Alternatively, the ultimate witness could be the sixteenth president of the United States, Abraham Lincoln, recreated virtually and drawing on more than 40,000 documents from the Lincoln Papers held at the Library of Congress to answer people’s questions. Pessach and Shur-Ofry also noted that the virtual eyewitnesses currently seen through hologram technology in some museums could, in the near future, become capable of interacting naturally with human beings. For example, these virtual witnesses could learn to respond to vocal cues indicating the age of the person asking a question and tailor their answers accordingly.

So, what if artificial intelligence were to create an eyewitness from the first generation of the Nakba who preserved all the oral testimonies of what the Zionist militias did in Palestinian cities and villages in 1948? Or what if it recreated the eyewitnesses who witnessed the massacres of Deir Yassin, Tantura, and others, as well as the families expelled from western Jerusalem, which was completely Judaized? Could the grandmother in every Palestinian home remain alive and tell, forever, the story of the forced displacement she endured? In fact, there is nothing that inherently prevents this from happening. The creators of New Dimensions in Testimony themselves note in their description of the program on their website that it is also designed for eyewitnesses and victims of other genocides, in addition to the Holocaust. Nor is there anything to prevent Palestinians, in the future, from seizing this opportunity and catching up with developments in artificial intelligence in order to begin recreating the first generation of the Nakba.

Fourth: Challenges to “Collective Memory” in the Age of Artificial Intelligence

The possibilities that artificial intelligence has opened up for collective memory are accompanied by challenges that AI has also created. Two of the most significant challenges have been identified.[11] The first concerns the nature of digital memory itself: it is not a neutral account of past events, and the memories conveyed by virtual witnesses are no different. Rather, these testimonies reflect the editorial policies established by those who operate the program. The second challenge lies in the ability of virtual memory agents and other AI-based technologies to be hacked or manipulated by third parties, or to suffer technical failures that could undermine the credibility of the mediated content. These challenges have attracted considerable attention from researchers in general, including Mykola Makhortykh,[12] who has identified four challenges that AI memory will bring with it. Among the most significant is the complex and opaque nature of algorithmic systems, which makes it difficult to detect malfunction or manipulation of collective memory. Indeed, this very opacity can facilitate the manipulation of memory by allowing the presentation of information that is incoherent, unsupported by reliable sources, or offensive to certain communities. Makhortykh has noted that some of these outputs may mislead users and alter their understanding of the past, while also enabling the “amplification of denialist content.” He discusses these challenges while warning that collective memory could potentially create greater opportunities for Holocaust denial. The same possibility applies to Palestinians and could likewise increase opportunities for denying the Nakba.

These challenges are not very different from those produced by the rise of digital platforms. Palestinian content has been subjected to deletion and restrictions by the major corporations that own social media platforms. Will the age of artificial intelligence reproduce these same challenges for Palestinians? Will Palestinian collective memory face the same forms of manipulation and restriction? And if Palestinians are able to create digital nomadic memories in virtual space—memories that travel across space and time and intersect with global campaigns—without becoming the kind of manipulated memory discussed by the French philosopher Paul Ricœur in his renowned book Memory, History, Forgetting, what mechanisms will they develop to preserve their memory within the realm of artificial intelligence?

Fifth: Memory Fiduciaries in the Age of Artificial Intelligence

Researchers Guy Pessach and Michal Shur-Ofry proposed a new framework that they call “Memory Fiduciaries” to address the challenges discussed above, as well as the ethical challenges associated with the use of artificial intelligence programs. They borrowed this framework from a similar one developed by Jack Balkin, namely “information fiduciaries,” which he discussed in the context of privacy concerns surrounding the collection of personal data by private companies. The researchers identified four duties of memory fiduciaries: identifying the bot; explainability; providing reasonable data security; and integrity. The latter requires fiduciaries to ensure accuracy and that the memories produced remain consistent with historical facts. But would this framework apply to the Palestinian case, particularly given that the proposed framework makes the operators of AI programs themselves the memory fiduciaries, on the premise that their relationship with users is based on trust, much like the trust that social media platforms have acquired? It appears, however, that this framework grants greater authority to the operators and assumes that adherence to these four principles qualifies a company or other entity to serve as a memory fiduciary.

Sixth: The Role of Memory Institutions, Such as Museums and Archival Centers, vis-à-vis For-Profit Companies

The concept of “memory fiduciaries” overlooks the fact that the entities operating AI programs are not always nonprofit organizations, such as institutions, museums, or archival centers. As AI programs continue to develop, the influence of private technology companies over collective memory appears likely to increase. Google Arts & Culture provides one example: the platform collaborates with more than 1,000 museums and cultural institutions and offers digital exhibitions on subjects central to collective memory, ranging from the fall of the Berlin Wall to apartheid. This brings us to the question of what the future holds for traditional memory institutions, such as museums, archives, and national libraries, particularly Palestinian ones. What we know at present is that, despite the enormous possibilities that programs developed by major corporations will offer for working with vast bodies of archival and historical data, the price may be “high,” as Wulf Kansteiner[13] has argued, because such collective memory will be “designed to serve the interests of its commercial owners.” In the Palestinian case, it could perhaps also be designed to serve the interests of the Israeli occupation and to ensure the continued dominance of its narrative at the expense of the Palestinian historical account. In a quick search of this website, I entered the term “Nakba” but obtained no results. I then searched for it under the historical events section and found only nine images listed under the heading “1948 Arab-Israeli War.” I also found a small collection of other images under the heading “1947–1948 Civil War in Mandatory Palestine.” By contrast, searching for “Holocaust” in the designated search field yields 13 collections of images, in addition to stories, museums, and exhibitions dedicated to the Holocaust. Are these the first indications of erasure in the age of artificial intelligence?

Related Books:

The Reign of Artificial Intelligence: A Heated Debate Between a Robot and a Human

Palestine: A Four Thousand Year History