Global AI Narratives: Latin America
In November and December 2020 the Leverhulme Centre for the Future of Intelligence (CFI) at the University of Cambridge joined forces with the Center for Chilean Literature Studies at The Pontificia Universidad Católica de Chile to conduct a series of three workshops devoted to AI Narratives in Latin America. Originally planned as an in-person, two-day event in April 2020 which was postponed due to the COVID-19 pandemic, GAIN Latin America shifted to a virtual format which showcased the work of more than twenty researchers, academics, practitioners and artists working in the field of AI narratives. Conducted almost entirely in Spanish, and attracting a large online audience for each event, these workshops highlighted a multiplicity of perspectives on artificial intelligence in Latin America today.
The first workshop, AI at the Edges of Latin America (27 November 2020), explored visions of a human-AI future from across Latin America and considered the role of Latin American peoples in the creation of AI narratives. The second workshop, AI in Digital Culture and Conceptual Approaches to AI (4 December 2020), considered the social implications of AI technology, but also highlighted the dislocation between public perceptions of AI and the ways AI actually works. The third and final workshop, Narratives and Poetics of AI in Latin America (18 December 2020), explored the situated nature of Latin American AI narratives and the ways in which they present some of the risks of AI. Across the three workshops the issue of inequality was raised a number of times - both within Latin American populations, and between Latin America and other world regions. Contributors highlighted the need to decolonise AI as key to ensuring a positive AI future for the peoples of Latin America, at both the individual and regional levels. This process should extend in a range of intersectional directions to incorporate not just geographic contexts, but also gender and other identities and demographics within Latin America.
Workshop I: AI at the Edges of Latin America (27 November 2020)
Professor Ignacio Sánchez, Rector of the Pontificia Universidad Católica de Chile, opened the first workshop which focused on AI at the edges of Latin America. Each of the presentations in this workshop highlighted in different ways the fluidity of the relationship between humans and AI - and asked what this relationship should look like in the future. Celebrated science fiction writer Alberto Chimal spoke about the idea of ‘captured intelligences’, arguing that science fiction in Latin America frames people as passive users, or even victims, of technology rather than active users or creators. Whilst Chimal considered how Latin American narratives frame the relationship between humans and AI, artist Raúl Cruz focused on the creation of these narratives, suggesting that Latin American peoples have a unique relationship with AI and that they can (and do) use this as a platform for creating science fiction that specifically focuses on their future with this technology. He highlighted the capacity of the Latin American peoples to write these narratives - they do not look for input from other parts of the world to create them. This offered a unique perspective of the Latin American peoples being simultaneously victims of the technology itself and active creators of the narratives that frame how they think about this technology. Cruz also considered the possibilities offered by speculative visual arts, suggesting that Latin American cultures can be preserved by exploring their relationships with technology through these media. Soledad Véliz, psychologist and science fiction author, considered the representation of relationships between humans and AI in Latin American science fiction stories. Moving beyond common sex-bot tropes, Véliz framed the discourse on these relationships as one characterised by reliance - of both humans on AI, but equally of AI on humans. She further argued that the debate around AI is one that should consider AI as a child, for whom humanity collectively acts as parents that have a responsibility to ensure the child is taught the appropriate ethical and moral behaviour.
Workshop II: AI in Digital Culture and Conceptual Approaches to AI (4 December 2020)
On the 4th of December 2020, the second workshop featured two panels: AI in Digital Culture in Latin America considered the ways in which AI influences cultural and artistic practices in Latin America today, and Conceptual Approaches to AI focused on how people understand (or misunderstand) the capabilities of AI. Carolina Gaínza opened the first panel with a presentation which focused on the opportunities offered by digital literature in Latin America and explored the relationship between AI and imagination in this context. Claudio Celis discussed AI-generated images and videos, including the development of so-called ‘deepfakes’, and explored the extent to which these types of image generation can be considered a form of machine vision. Martín Tironi looked towards the future of AI, and discussed likely versus possible future trends and risks in AI, whilst strongly emphasising the need to decolonise AI design, development, and implementation. Claudia López argued that the decolonisation of AI and the promotion of ethical AI development and use would inherently require a shift towards incorporating co-creation or co-design techniques in AI development to ensure that those people most affected by the use of AI have the chance to have their voice heard in its development. Wolfgang Bongers, in a talk that linked Claudio Celis’ reflection on AI creating art with Raúl Cruz’s perspective on incorporating technology in art, discussed the possibilities and limitations of AI-generated art and asked us to consider what the act of creation really requires. Prompted by an audience question, the panelists discussed the possibilities of AI being used to target white-collar (as opposed to violent) crime in Latin America, raising questions about surveillance and the targeted use of AI, in particular against demographics such as those of lower socio-economic status and people of colour.
Felipe Cussen opened the Conceptual Approaches to AI panel. He discussed the concept of artificial stupidity and the possibility of AI systems amplifying errors because the systems lack the ability to identify when their solutions are absurdly wrong. He suggested that it is an AI’s lack of self-awareness of its own limitations that is potentially more dangerous than simply coming to the wrong conclusions - he demonstrated this point by showing that through minor pixelation of an image, image recognition software can quickly misidentify an image of a person as a rug! Mario Ponce then explored how people think about how AI works, demonstrating that a lack of understanding of some simple statistical methods can lead to AI being perceived as almost magically omniscient, with similar techniques promoting widespread belief in a high level of surveillance by social media companies. Tomás Vivanco talked about how using AI techniques can revolutionise the use of bioplastics in remote communities, exploring the possibilities of creating bioplastic ‘recipes’ for the creation of building materials in remote areas and using AI algorithms to identify the most efficient way of utilising these materials. Interdisciplinary artistic duo Montenegro Fisher closed the panel with a powerful combined spoken word and visual performance which explored how AI might reframe how we consider the experience of Latin American Indigenous peoples, but also how AI might perpetuate misrepresentations of the experiences of these peoples through an assumption of contemporaneity. The discussion from this panel focused around ideas of techno-failure, with both Felipe Cussen and Montenegro Fisher arguing that they (in different ways) use this possibility of techno-failure as an opportunity to reframe and reconsider the possibilities of AI.
Workshop III: Narratives and Poetics of AI in Latin America (18 December 2020)
The third and final workshop considered narratives and poetics of AI in Latin America. Macarena Areco opened with an exploration of the ways in which AI is imagined in Latin American science fiction, and how these imaginaries reflect tensions between differing ideas about what a technological future may look like. Joaquín Jiménez discussed Peruvian author Santiago Roncagliolo’s visions of a capitalist dystopia in his 2010 novel So Close to Life, which provided a unique insight into the Latin American view looking out at other regions in relation to AI, rather than exploring how other parts of the world look in at Latin America. Sofía Rosa discussed possibilities (both positive and negative) of becoming reliant on digital interfaces for human-to-human interaction that are presented in Uruguayan author Gabriel Peveroni’s 2004 novel El exilio según Nicolás (The exile according to Nicolás) - a surprisingly relevant novel given its premise of a deadly disease sweeping the earth and drastically changing how people live their lives. Richard Parker closed the panel with the presentation of powerful spoken word poetry which explored themes of identity, technology and place in Latin America. In the roundtable discussion, Kanta Dihal, Wolfgang Bongers, Ramiro Sanchiz and Andrés Tello addressed ideas about the whiteness of AI and the need to decolonise the most prominent representations of AI - representations which are often imported to Latin America from other parts of the world. Tello argued that Latin American peoples accepting and using AI in its current forms, as developed in other parts of the world, would mean accepting a new wave of digital colonialism.
Featuring a diversity of speakers from across Latin America, from Mexico to the Antarctic regions, this workshop highlighted a multiplicity of perspectives. However, key themes arose that show some clear convergences in current thinking about AI in Latin America. The workshops considered what it means to be a user of AI technology in Latin America, and the uniqueness of this experience in an ever more globalised world. It also considered how imaginaries of AI transgress boundaries; how they enter Latin America from other regions, and also how Latin American visions might transgress borders, asking what influence these specific ideas might have on a global scale. The concept of AI as linked with colonial practices was discussed from a number of different perspectives, with the decolonisation of AI highlighted as an important strategy in developing AI that works for Latin America. This discussion around decolonisation went beyond simply the geographic, and sought to consider other demographics, including gender and race. Overall, this workshop highlighted a substantial and unique body of thought on AI relevant to the experience of Latin American peoples today.
The Global AI Narratives (GAIN) project is an interdisciplinary, cross-cultural initiative seeking to understand local interpretations of artificial intelligence (AI) in different countries and communities around the world. The project is funded by the Templeton World Charity Foundation and DeepMind Ethics and Society. The Center for Chilean Literature Studies is an interdisciplinary research center dedicated to the study, preservation, and diffusion of Chilean writing and culture.
The Global AI Narratives and Center for Chilean Literature Studies teams would like to thank all of the wonderful speakers, participants and audience members for their contributions to this series of workshops.
Image courtesy of Raúl Cruz