Historical critiques

When historicizing current AI developments, it is common to refer to earlier phases of AI growth and stasis as AI ‘summers’ and ‘winters’ (Haigh Reference Haigh2023; Markelius et al. Reference Markelius, Wright, Kuiper, Delille and Kuo2024). However, this reinforces the narrative promoted by big tech of (interrupted) technological progress that will inevitably lead to so-called ‘general’ or ‘strong’ AI. Historical critiques complicate this seasonally inflected narrative by exploring genAI as the outcome of interconnected technological, social, cultural, economic, and political processes (as indicated more in later sections of this commentary) and emphasizing the views of those who narrate AI’s history differently.

Common histories of AI often begin with the decontextualization of Alan Turing’s oft-quoted question, ‘can machines think?’, shaping present (mis)understandings of genAI (Turing Reference Turing1950; Proudfoot Reference Proudfoot2011). Turing was interested in imitation and distinguishing between ‘discrete-stage’ (computers) and ‘continuous-stage’ (humans) machines. The Turing test that gives ‘computer scientists a sense of direction’ (Stilgoe Reference Stilgoe2023) is, thus, often misremembered as seeking to develop machines that think humanly rather than machines that mimic human thinking, foregrounding the deceitfulness of genAI (see Natale Reference Natale2021).

At the 1956 Dartmouth Summer Research Project on AI, the pursuit of general/strong AI and the idea that all human intelligence could be ‘so precisely described that a machine can be made to simulate it’ took further hold (McCarthy et al. Reference McCarthy, Minsky, Rochester and Shannon2006, 12). Some of its participants were later critical of this view (Minsky Reference Minsky1986) and later leading voices in the field like Joseph Weizenbaum – creator of ELIZA, commonly considered the first chatbot (developed in 1964–67) – criticized early AI boosterism as conservatively promoting technical solutions that left existing power hierarchies intact (Reference Weizenbaum1976; Birhane et al. Reference Birhane, Ruane, Laurent, Brown, Flowers, Ventresque and Dancy2022). More recently the ‘mathe-morphized’ historical narrative of AI that equates the ‘precise description’ of intelligence with ‘mathematical description’ and prioritizes the pursuit of general/strong AI has been further problematized by research on AI, race and indigenous knowledge systems that stresses the existence of multiple intelligences and subjecthoods (Buolamwini Reference Buolamwini2023; Lewis et al. Reference Lewis, Whaanga and Yolgörmez2025; Richardson-Walden and Makhortykh Reference Richardson-Walden and Makhortykh2024).

In complicating notions of a ‘generalised’ intelligence that can be modelled, such research offers alternative perspectives which may better serve and represent the diversity of human, social, and cultural memory, and their possible interfaces with computer systems. These critiques sometimes still imply that LLMs are comprehensive and have autonomous agency (Richardson-Walden and Makhortykh Reference Richardson-Walden and Makhortykh2024), but they also remind us that the rationalist, mathematical way of ordering things that has dominated Western thought since the 18th century (see Foucault Reference Foucault1994) is not the only way. As Lewis et al. (Reference Lewis, Whaanga and Yolgörmez2025) argue, indigenous knowledge systems foreground intelligence as a collectively established relationality rather than a property of isolated individuals, making visible the social and cultural matrices in which notions of ‘intelligence’ emerge and acquire value.

GenAI might then be approached as narrow/weak AI that gains value by posing as general/strong AI (via anthropomorphization) even as its ongoing advancement and integration with other forms of AI complicate this. As part of the future projection of the AI narrative of progress, genAI is frequently credited with excessive levels of agency, reversing Latour’s observation that humans typically attribute limited agency to machines (Reference Latour1987; see Smits and Wevers Reference Smits and Wevers2022). Still, many earlier technologies have had profound effects on memory, and thus it is important that genAI’s mnemonic consequences are sufficiently historicized by interrogating their similarities and differences from older examples of technology-assisted remembrance. While memory scholars have historically conceived technology in a human-centric and instrumentalized manner, opening to wider histories that challenge anthropocentric and anthropomorphized historical accounts that abstract genAI can arguably help us better understand how it contributes to social processes of remembering within which agency is always distributed between human and non-human actors (Merrill Reference Merrill and Lindgren2023; Smit et al. Reference Smit, Smits and Merrill2024).

In short, memory scholars are well-positioned to problematize hype-driven narratives of general/strong AI’s inevitability. They can ask what mnemonic shadows are cast by AI’s dominant historical narratives, how did these narratives emerge and proliferate, and what can be gained by re-reading them against the grain? Here, memory studies scholars can learn from those working in critical AI studies and in media philosophy who contextualize genAI technologies within longer media debates (Natale Reference Natale2021; Lindgren Reference Lindgren2024) and encourage us to consider to what extent the ontologies, epistemologies, and aesthetics underpinning AI are actually radically new (Fazi Reference Fazi2019, Reference Fazi2024). They can also draw on interdisciplinary perspectives by, for instance, combining science and technology studies, literary studies and media studies to reveal how genAI’s accepted historical narratives are linked to the exercise of social and political power in the present (Cave et al. Reference Cave, Dihal and Dillon2024; Magalhães and Smit Reference Magalhães and Smit2025). Alternatively, the analysis of oral histories, memoirs, and autobiographies could provide insight into the memory of AI’s development via the experiences of those directly involved in it. Media archaeological and historical computational science approaches could also be added to the interdisciplinary mix to re-read AI development via, for example, the resurrection and reinterpretation of old and now obsolete computer code (see Kilgrove Reference Kilgrove2025). Ultimately, problematizing the AI hype to understand the relationship between memory and genAI from a historical perspective requires us to be technologically informed, without being technologically deterministic. This leads us to the next area of critique.

Technical critiques

To understand genAI’s implications for memory, it helps to know how it functions. This area of critique problematizes singular, diffuse notions of genAI by stressing the importance of ‘deblackboxing’ (Dixon et al. Reference Dixon, Hsi and Oh2022). This is the effort to make the processes of not only computation but computing generally (i.e. including the effects of sociotechnical infrastructures and power relations on technology) more transparent. This is not straightforward because, as Crawford and Joler (Reference Crawford and Joler2018) have ‘anatomically’ captured, the complexity and scale of genAI systems almost exceeds human imagination in relying on a vast (and rapidly changing) capitalistic matrix of hardware and software and human and non-human relations. On the technical level, this means that it is difficult to explain why (especially more complex) AI systems make concrete decisions even under conditions of full transparency because transparency does not automatically lead to comprehension (Esposito Reference Esposito2022). One way to pursue ‘deblackboxing’ that might be helpful for memory studies researchers is to turn to the computer science texts in which (basic) AI principles and procedures are described and theorized. Exemplifying this, Amoore et al. (Reference Amoore, Campolo, Jacobsen and Rella2023, 1) approach these texts as contested sites ‘through which machine learning shapes the world’. For memory studies scholars, such an approach can help reveal how computer scientists understand and approach memory and the processes of remembering and forgetting in computers and software (see Merrill Reference Merrill and Lindgren2023).

Still, genAI systems can generally be technically understood as involving the three pillars of big datasets, high-performance computing infrastructures, and the machine learning algorithms that are applied to data to synthesize content, all underpinned by computer science and mathematics (Van Assen et al. Reference Van Assen, Banerjee and De Cecco2020). For instance, chatbots, or by their less anthropomorphized name, transformer-based instruction-tuned text generators, are guided by computational training tasks and mathematical principles (e.g. probabilistic reasoning) that are applied to large volumes of training data to create models powering interfaces that respond to user inputs in line with specific patterns in the training data (Smit et al. Reference Smit, Smits and Merrill2024; see Paglen and Downey Reference Paglen and Downey2023 on genAI image creators).

An important principle for many genAI applications is to model a baseline (Chen and Chen Reference Chen and Chen2022). For the LLMs behind chatbots, this baseline is conventionally equated to the statistical prediction of units called tokens (e.g. words, letters, and symbols), based on their training data set–mimicking language patterns but without understanding what they are ‘chatting’ about (Bender et al. Reference Bender, Gebru, McMillan-Major and Shmitchell2021). GenAI systems are, thus, profoundly dependent on their training data, and the computational, and often implicitly cultural, principles prioritized by their designers and developers, including how to approach outliers, parameterization, and randomization. These principles are rarely divulged publicly. Similarly opaque is how these principles are translated into ‘guardrails’ that aim to prevent the misuse of genAI. So too, the training data behind different chatbots, including whether it is original or synthetic (e.g. itself AI-generated), is often a closely guarded secret, even if this determines the semantics of the content produced, the risks for so-called ‘model collapse’, and ethical concerns about data ownership and privacy. Critically, it is also unclear how far designers and developers take into consideration the social impacts of their decisions as their hypothetical ‘user’ becomes millions of users (Salvaggio Reference Salvaggio2025). While some of these technical critiques apply to other technologies also, collectively they raise important questions when considering genAI and memory’s relationship.

Most fundamentally, we might ask: should genAI be designed to only create factually accurate content, or should it allow greater degrees of mnemonic creativity? Should models forget and machines unlearn (cf. Bourtoule et al. Reference Bourtoule, Chandrasekaran, Choquette-Choo, Jia, Travers, Zhang, Lie and Papernot2021)? What might it mean for memory when future outcomes are statistically modelled on past evidence? How far, then, might relinquishing memory to statistical, rather than cultural, weighting create hegemonic, ‘line of best fit’ forms of remembrance that further diminish the importance of mnemonic outliers? What, in short, are the memory baselines of genAI? Are they based on official historiographies or a greater diversity of sources, and how should the inevitable differences between these be addressed? Are memories that do not match an agreed baseline or fall outside its parameters no longer valid? What might this mean for the contestation of memory and phenomena such as memory activism? Such questions also have implications for the design of genAI systems specifically for mnemonic purposes, whether chatbots in heritage institutions or personalized digital duplicates of historical figures (see Kozlovski and Makhortykh Reference Kozlovski and Makhortykh2025). These and other mnemonic uses should determine the computational logic behind genAI applications, the training data required to implement them, and the guardrails preventing their possible misuse. It encourages memory scholars to be involved in these design processes and decisions, but also discussions about genAI policy, regulation, and law. We consider what all this may mean for users, for example, in terms of who decides on and differentiates between appropriate and inappropriate mnemonic uses of genAI, in the next section dedicated to praxis critiques.

Praxis critiques

A key debate regarding the use of genAI in everyday life relates to whether it will help extend our memory, cognitive capacities, and creativity. Or, alternatively, whether we will offload our memory and knowledge to genAI so much that it compromises our intellectual autonomy. Navigating this debate, which has long characterized the subfield of digital memory studies (Hoskins Reference Hoskins2011, Reference Hoskins2013) depends, in part, on understanding users’ differing levels of expertise and AI literacy (Imundo et al. Reference Imundo, Watanabe, Potter, Gong, Arner and McNamara2024).

Experts in different fields use genAI to acquire synthesized information and feed creative thinking (Javaid et al. Reference Javaid, Haleem and Singh2023; Zhu et al. Reference Zhu, Gao, Wang, Liao, Zheng, Liang, Wang, Pan, Harrison and Ma2024). Their specialized knowledge also allows them to better detect incorrect or incomplete genAI outputs related to their field, although they are not immune to genAI’s errors, especially if they rely on it as an external memory aid (Fisher and Oppenheimer Reference Fisher and Oppenheimer2021; Azaria et al. Reference Azaria, Azoulay and Reches2025). Novices meanwhile may use genAI (e.g. chatbots) to learn because it provides accessible, well-organized, and coherent information, through human-like dialogues, but the lack of specialized knowledge makes novices more vulnerable to errors in the generated content (Fang et al. Reference Fang, Lippert, Cai, Hu, Graesser, Sottilare and Schwarz2019; Hennekeuser et al. Reference Hennekeuser, Vaziri, Golchinfar, Schreiber and Stevens2024). No matter then what a user’s level of domain expertise may be, expertise in using AI or AI literacy is also important.

There is a need for more research on how genAI–human interactions influence human memory capacities and what boundary conditions shape this process. What might be the optimal conditions that unburden human memory through offloading to genAI while maintaining a critical knowledge base when partnering with genAI? How might the discrepancies in human mnemonic capacity that can be both diminished and exaggerated by genAI relate to expertise, but also socioeconomic, racial, gender, or cultural differences? The role of memory scholars here could expand beyond studying genAI’s praxis-related implications to becoming ‘domain experts’ that can co-design AI and influence its surrounding policy and regulation towards cognitively advantageous and societally just outcomes.

Recognizing the impact of differential expertise also connects to debates about whether genAI enables or endangers human mnemonic agency. Does the ability of genAI to synthesize ‘the past’ deprive our memory of authenticity and render our life stories anti-autobiographical (Hoskins Reference Hoskins2024)? Or do individuals and groups retain power in making decisions and choices over what to remember and what to forget despite the technological hype (Wang Reference Wang2019)?

The urgency of these questions is stressed by genAI ‘chatbots’ possessing the illusionary appearance of human features that seem to reduce or replace human agency. They can assume various human-like personas, perform highly on tasks that require sensitivity to human emotions, and provide instant insights on complex intellectual questions via dialogues that reinforce the illusion that they possess human-like consciousness (Elyoseph et al. Reference Elyoseph, Hadar-Shoval, Asraf and Lvovsky2023; White et al. Reference White, Fu, Hays, Sandborn, Olea, Gilbert, Elnashar, Spencer-Smith and Schmidt2023). Chatbots can also automatically acquire and generate information about individual and collective pasts without human approval or control (Hoskins Reference Hoskins2024) and shape the mnemonic agency of human users according to how they have been designed and developed, for example, in terms of their parameters of possible interaction.

Designers and developers, as humans, command their own forms of mnemonic agency (Smit et al. Reference Smit, Smits and Merrill2024). Humans also retain agency, in addition to their expertise or AI literacy. They can decide when, how, under what circumstance, and for what purpose to use genAI, just as when they confront other digital technologies (Wang Reference Wang2019). They often provide both the prompts and the data – typically in the form of information about their individual and collective pasts shared on social media (see Wang and Hoskins Reference Wang and Hoskins2024 ) – which are then used to train LLMs and ultimately shape what chatbots ‘reassemble’ for them as ‘memories’. Neither is their remembering only restricted to the content that these chatbots provide them – context is also important with humans and genAI prompting each other to remember (Smit et al. Reference Smit, Smits and Merrill2024). Historicizing genAI, this process is not unlike the transactive, dialogical, and phenomenological constructions of one’s autobiographical memory that occur within other online settings (Wang Reference Wang2022; Merrill Reference Merrill, Ekelund, Guðmundsdóttir, Merrill and Sindbæk Andersenforthcoming). It is thus critical that memory scholars work to disentangle how human mnemonic agency interacts with that of genAI and avoid totalizing and sensationalist prognoses of the loss of human agency, which contributes to the AI hype. Indeed, there are good arguments for thinking of the ‘A’ in ‘AI’ differently – in terms of augmentation rather than artificiality (Dekeyser and Whitehead Reference Dekeyser and Whitehead2025).

Geopolitical critiques

Beyond problematizing how users from various global demographics may remember differently with genAI, applying a geopolitical critique highlights how global power dynamics – particularly involving the EU, US, and China – shape the development and distribution of genAI (Larsen Reference Larsen2022; Kennedy Reference Kennedy2025) and, in turn, the global battle to control public narratives regarding historical and contemporary events. With AI historically embedded in capitalist logics and driven by government funding priorities and military support for scientific research (Nilsson Reference Nilsson2010; see also Pilkington Reference Pilkington2024), the current hype can thus be understood in the context of major geopolitical and economic uncertainty and AI’s increasing use in warfare. Indeed, the companies behind the most prominent commercial ‘chatbots’ are now pivoting towards military contracts (O’ Donnell Reference O’ Donnell2024).

The competition for global genAI dominance and thus political and mnemonic influence has been intensified by the rising economic, technological, military, and political power of China in recent decades. Although the US still holds an edge in advanced AI systems, China is catching up quickly through the development of open-source LLMs, strategic investments, and government support (Kennedy Reference Kennedy2025). Pertaining to memory, this genAI geopolitical rivalry will likely increasingly contribute to and coalesce with globally polarizing debates regarding historical nihilism and historical revisionism. Relatedly, there is recent evidence of US-owned genAI ‘chatbots’ being used to promote Russian geopolitical interests by aiding the censorship of undesired pasts (Urman and Makhortykh Reference Urman and Makhortykh2025).

Global South countries, including in Africa, Latin America, and Southeast Asia, also play an active role in genAI geopolitics through alliances and by setting regulations (see Feakin Reference Feakin2025), but a geopolitical perspective also highlights the uneven global distribution of genAI’s costs and benefits between the Global North and South. Many ‘chatbots’, for example, work discriminatorily by silencing, and under- or misrepresenting different minority groups across the world (Okolo Reference Okolo and Lindgren2023). This perpetuates and amplifies unequal global power relations, further disempowering already marginalized communities and countries. The size of LLM training datasets does not guarantee their diversity (Bender et al. Reference Bender, Gebru, McMillan-Major and Shmitchell2021). At every stage of their curation – from initial online participation to data collection and fine-tuning, including reinforcement learning with human feedback, current practices favour hegemonic perspectives (Smit et al. Reference Smit, Smits and Merrill2024). Even while genAI can have democratizing mnemonic effects at certain scales, in general, memories of certain hegemonic groups are widely represented, while those related to marginalized groups are underrepresented or absent. As such, the datafied memory of Global South cultures is often missing due to ‘algorithmic exclusion’ (Albert and Delano Reference Albert, Delano and Lindgren2023), ‘digital cultural colonialism’ (Kizhner et al. Reference Kizhner, Terras, Rumyantsev, Khokhlova, Demeshkova, Rudov and Afanasieva2021) and other processes of digital suppression, including, in some contexts, state-led censorship.

GenAI’s under-representation of non-hegemonic memory is partly linked to the dominance of English in computational linguistics (see Joshi et al. Reference Joshi, Santy, Budhiraja, Bali and Choudhury2020; Bender et al. Reference Bender, Gebru, McMillan-Major and Shmitchell2021). GenAI, thus, can often perpetuate colonial knowledge regimes that disregard alternative ways of understanding and interpreting the world (Birhane and Talat Reference Birhane, Talat and Lindgren2023; Lewis et al. Reference Lewis, Whaanga and Yolgörmez2025). In turn, genAI’s ‘average collective memory’, its mnemonic ‘line of best fit’, can hide a richer diversity of remembrance cultures (Makhortykh Reference Makhortykh2024; Schuh Reference Schuh2024). This process not only reinforces existing power imbalances regarding which memories can be accessed and which cannot – new forms of memory imperialism – but also erases the distinct meanings that memory, forgetting, or trauma may hold for marginalized communities. Meanwhile, the rapid emergence of LLMs in high-resource languages like Chinese and Russian (e.g. CT-LLM or YandexGPT) but also low-resource languages like Kazakh and Swahili (e.g. KazLLM or UlizaLlama) offers alternatives that may foreground other renderings of the past.

GenAI systems also rely on an uneven international division of digital labour (Fuchs Reference Fuchs2014). They are designed and owned by powerful companies in the Global North, but various stages of their implementation are outsourced to the Global South. A significant part of the data collection pipeline – data labelling – relies, for instance, on ‘ghost workers’ in the latter. While this labour helps generate massive earnings, the profits are captured by others, creating a stark disparity between the millions earned by data labelling companies and the low wages of their workers (Okolo Reference Okolo and Lindgren2023). Similarly, ‘chatbots’ have been tested in refugee camps without adhering to the rigorous ethical procedures applied elsewhere. These practices rework and revitalize ‘colonial genealogies through processes of extraction, coloniality, control, and discrimination’ (Madianou Reference Madianou2025, 18).

Questioning and overcoming these global mnemonic power imbalances should lie at the heart of memory studies’ genAI-related concerns. We need to ask: will global competition for AI dominance benefit end users by creating more divergent narratives, views, and perspectives or polarize the world even further? Whose memories are being preserved, transformed, or erased by genAI? Can genAI accommodate the plurality of mnemonic epistemologies across cultures, or are they only able to reproduce hegemonic views? How might memory scholars work alongside communities affected by digital erasure to resist genAI’s technocolonial logics? Can genAI systems be reimagined not as tools of erasure but as platforms for restorative memory work – and if so, under what ethical and political conditions? Such questions are especially important given that those communities least likely to benefit from genAI are also often the most vulnerable to its harms, especially those of an environmental character.

Environmental critiques

Environmental critiques of AI and its surrounding hype complicate claims that digital technology, through its specific temporalities, has radically reformulated human remembering and forgetting, and that this is exacerbated by genAI (Ernst Reference Ernst2013; Hoskins Reference Hoskins2013; Hoskins Reference Hoskins2024). Whilst notions of decay time and entanglement have been crucial in digital memory studies, this has predominantly focused on acknowledging the relationality of self and machine, and self and data (Hoskins Reference Hoskins2013, Reference Hoskins, Hajek, Lohmeier and Pentzold2015, Reference Hoskins2024). Widening focus to emphasize the wider environment, brings the discussion of AI’s mnemonic implications and, in turn, the subfield of digital memory studies into the realm of ‘fourth wave’ memory studies research (see Erll Reference Erll2024). Expanding focus to the broader geological plane of ‘media’ (to include all meaning-making matter), attention is drawn to planetary decay happening at alarming rates, accelerated by an increasing dependence on digital technology (Parikka Reference Parikka2015; Crawford Reference Crawford2021).

This focus resonates with memory studies’ ‘anthropocentric turn’ and the growing consideration of ‘planetary memory’ (Bond et al. Reference Bond, Rapson and de Bruyn2018), the ‘deep-time of the earth’ (Chakrabarty Reference Chakrabarty2009), ‘terrestrial memory’ (Golańska Reference Golańska2023), and attention to those human and non-human relations in ways that hold us ‘responsible and accountable for our actions towards’ the other (Kennedy Reference Kennedy, Heise, Christensen and Niemann2017, 506). Zooming out further, it brings the wider organic ecosystems of our planet (and beyond) into view. To resist the genAI hype through an environmental critique, then, is to expand the spatial and temporal dimensions of our focus away from only interactions with, and the outputs produced by genAI, towards the entanglement of deep time ingrained in, yet hidden by, the convenient instancy of genAI interfaces. As Reading argues, we need to go further, beneath and beyond the surface level of ‘the skin or screen of digital memory’ (Reference Reading2014, 753; see also Loots Reference Loots2024). This goes even further than tracing the capitalistic technological infrastructures and human relations on which genAI relies, in also foregrounding the actors involved in the production of genAI in terms of resources and energy consumption, and the long-term consequences of extraction and power use.

Exemplifying this, Crawford and Joler’s ‘anatomy of AI’ exceeds the technical in acknowledging that whilst our ‘encounters with AI are fleeting and brief’, behind each lies the ‘interlaced chains of resource extraction, human labour and algorithmic processing across networks of mining, logistics, distribution, prediction and optimization’ (Reference Crawford and Joler2018). Thus, whilst typing a quick prompt into a chatbot might seem frivolous, the energy use that enables computation and the material resources required to create, maintain, and extend the hardware necessary for ever-expanding data storage and processing are disruptive to existing ecosystems. As Crawford elsewhere notes, ‘from the perspective of deep time, we are extracting Earth’s geological history to serve a split second of contemporary technology time’ (Reference Crawford2021, 31).

In principle, there is nothing new here. Media – digital or otherwise – have always relied on substantial material extraction, production, and waste (Maxwell and Miller Reference Maxwell and Miller2012; Parikka Reference Parikka2015), and the tension between the illusion of immateriality (e.g. the ‘cloud’, the AI ‘black box’) and this continuous geological deep time of media is built into the algorithmic logic of computing. Yet, the mainstreaming of genAI, which promises forms of enhanced computing, and working at complexities, scales, and speeds beyond human capacities, makes us feel unintelligent, irrelevant, and somewhat powerless in its shadow. Meanwhile, these systems propel the destruction of our ecosystems irreparably, leaving their imprint on the planet eternally.

The field of memory studies has paid relatively little attention to the broader material consequences of the ‘often obfuscated environmental exploitation and friction between capital and labour that go into these newer forms of mediated memory’ (Reading Reference Reading2014, 749; see Loots Reference Loots2024). Adopting an environmental critique to genAI calls on us as memory scholars to go beyond the interface encounter and the illusion and deception of this experience (Natale Reference Natale2021). How can we be attentive to what is not visible in that moment? This demands an infrastructural approach to memory construction; that is, there is a need to scrutinize how current AI-enabled memory practices and technologies are materially supported. Memory scholars could make visible how our planetary and supra-planetary environments and resources – the deep time of our planet and universe – are problematically entangled with our sociotechnical systems of memory. A focus on the environmental dimension of genAI, thus, helps shift our critical gaze towards the memory of the Earth and its universal neighbours. Ultimately, this focus offers a posthuman approach to the entanglements of humans and nonhumans in producing genAI and creating memory. This ‘holds the potential to cultivate response-able forms of memory, reshaping how essential interdependence is practiced in the everyday rituals of living and remembering within our more-than-human world’ (Gündoğan İbrişim Reference Gündoğan İbrişim2024, 101).