Hearing voices and imagining the rainforest: Why Artificial Intelligence (<scp>AI</scp>) cannot compete with the need for the ‘human touch’ in research analysis

Women's Voices from the Rainforest. Janet Gabriel Townsend (in collaboration with Ursula Arrevillaga, Jennie Bain, Socorro Cancino, Susan F. Frenk, Silvana Pacheco and Elia Pérez). Routledge, London, UK and New York, USA, 1995, pp. viii + 212. ISBN 978-0-415-10532-3 (pbk). After three decades of working in Latin America, and a year before the COVID-19 pandemic hit Peru so badly, I found myself in the Loreto jungle, several days up-river from Iquitos, flat out in the mud, wet from head to toe, and wishing for the first time that I was anything other than a geographer on fieldwork. After years of trying to channel Janet Townsend's politics of research into my own work, I was on the verge of failing. I felt too defeated to focus on the outstretched arms of my Peruvian colleagues readying to pull me up (yet again)—‘in-person’ proof of our much-valued collaboration. My frustration at the environment around me was gaining pace as the strength of my thumping heatstroke-headache grew with each sodden step. I had arrived in Loreto, my ears ringing with geography stories about the Amazonian jungle and of what it would reveal to us once we could start to analyse our hard-won data. I had witnessed first-hand my physical geography colleagues’ joy when their pollen samples under the microscope again verified the existence of a vast intact tropical peatland in the Pastaza-Marañón Foreland Basin. If kept in the ground, this peat would serve as an important carbon sink, giving some protection against climate change (Roucoux et al., 2017). Prior to my departure, reading my historical geography colleague's work on Pierre Gourou and tropicality made it clear that the idea of ‘the jungle’ as ‘terra nullius’ is a fiction of empire, aiding colonial domination since the invention of the ‘tropics’ (Bowd & Clayton, 2019). So, once my data collection was complete, my analysis would need to look for evidence of how the lives of people and ‘other than humans’ have been made invisible over time by colonialism and a capitalist intent of ‘progress’. Inspired by Janet Townsend, my mentor, close friend and colleague, it was the lives of women that I had come to Loreto to find out about. I wanted to hear their voices from the rainforest so that my analysis could be as genuine, reflexive and groundbreaking as hers had been. Face down in the Amazonian mud, Janet came to mind to remind me that there is nothing comfortable about research—it requires labour, emotional and physical. Our previous collaboration in Nepal with ‘Shakti Samuha’, an organization founded by women who have experienced trafficking, had been a harrowing lesson in realizing that dealing with mud is frequently the least of our problems (Laurie et al., 2015). My motivation to turn to a new field of research, to learn about ways to heal our planet and our human relationships with and within it, was, in part, because of how heartbreakingly hard our post-trafficking work had been. A constant thread running through all of Janet's previous research had been listening to women's voices as they talked about their past experiences in diverse development settings. Her approach encouraged them to imagine who and what they could become in the future, as they spoke their fears, hopes and dreams out loud. And here was Janet again, doing the same for me, beyond time and place and yet located, appropriately, in a rainforest, giving me the human touch. So, I took the hands that were offered, and with their help stood up. Women's Voices from the Rainforest (WVFTR), like its lead author, was far ahead of its time: frank non-judgemental treatment of women's dilemmas in balancing nature and conservation with economic and family needs; worked out and well-documented co-production practices underpinning work that stubbornly rejected relegating labels like ‘only’ action research, as if any human geography research should be without action; the attention to crediting authors and collaborators decades before journals introduced tables and categories for such things. Just like the women it presents to the reader, in all these ways this was a pioneering text. The book's power is that it speaks to us relationally, we see ourselves and our mothers and grandmothers in the stories that are told and how they are narrated. Alongside witnessing a new generation of students captured by this book in my teaching, it is the detail in explaining the analysis process that has most recently called my attention. What does the human touch of the kind that Janet and her collaborators set out to offer that new AI translation tools and forms of analysis cannot compete with? To begin to answer this question, I locate my comments within the recent debate about the poor state of analysis in qualitative research, and specifically Hitchings and Latham's (2020) critique of what we geographers say about our analysis process when we write and publish. I have built on their critique recently in my own work on El Niño in northern Peru (Laurie et al., 2024), but I have also questioned their lack of ‘explicit engagement with the feminist legacies in qualitative research’ (Laurie et al., 2023: 251), mentioning WVFTR in particular. The book was ahead of its time in terms of the detail it gives on how and why analysis of qualitative data collected across Mexico and Colombia, in different time periods, took the form that it did. It also raised hitherto ignored methodological questions about what such careful (and diverse) forms of collaborative data collection mean for how we read and present what we find. The book reveals this process as something lived and co-produced through human relations and human-nature relationality at specific life moments for women in the rainforest. It is this practice, which from my perspective as a human researcher, AI cannot replicate. It can give us something else, but it cannot co-produce in this way. Janet Townsend first read all the life histories straight through, noting which seemed to be discussed by many women at length and with interest. She then extracted comments on these themes and grouped them together, before adding comments on the same themes from our notes on unstructured interviews. The chapter was thus written around the women's words (p. 81). Most life histories were recorded by two of us and transcribed by one although some were the work of solely one person. We all participated in this task, which we found very satisfying, if sometimes painful (p. 55). WVFTR was groundbreaking in speaking directly and openly about how power differentials in data collection and analysis are not fixed but flow in multiple directions. As outsiders, opportunities arose through the research for Janet and her collaborators individually and as a team to be manipulated into specific narratives and agendas that could serve the community. Such entanglements replicate the everyday ways in which marginalized communities have been required to influence local political systems, to access what was traditionally theirs, since early colonial times. What would AI make of all this? Could we program parameters to highlight moments where this type of ‘capture’ might happen? Probably, but by default it would leave out the emotional labour, the gains and pains of the ‘in person’ exchange. Field encounters in WVFTR are described as both flawed and familiar. They are set up within the everyday forms of negotiation that ‘place identification’ and its associated personal histories of care structure. This is how the research opened a space for stories to be told and heard. Three decades on from its publication, I am reflecting on the legacy of the book for my current work with young people in northern Peru. Amidst dominant disaster narratives about the impact of El Niño in this region, they are becoming the storytellers in our research collaboration. Their intergenerational oral histories, video and music making chronicle the positive benefits of El Niño rains in their part of the desert (Laurie et al. 2023). Like the women pioneers in the rainforest, their voices have not been listened to before, in part because their story runs counter to accepted thinking (Bell et al., 2024). We all know that AI can write books, even novels and poetry, but it cannot tell a story like the rainforest voices do. Does all this mean that the methodological contribution of WVFTR was so far ahead of the curve that it was already paving the wave for what looks set to become the new ‘storytelling turn’ in geography? I tend towards thinking so. This was drafted with support from Leverhulme Trust: MRF-2022-065: ‘Development futures in a covid, climate challenged world: a decolonial approach’.