Be Black, Be Bold, Be Innovative, Show the World Equitable Geo. We're coming together as a collective to celebrate people of African descent, the diaspora, and talking about geospatial equity and justice. You're listening to The North Star Gaze, a podcast with intimate stories from geoluminaries. Hi everyone. My name is Yariwo, host for this episode. So when we think about tech, data science, machine learning, and GIS, we really think about it in context of history and humanities, and I think that's what really struck me about this particular conversation and this guest. We constantly keep saying GIS is limited by our imagination, but I think. Our guest embodies just that, because after I looked at the research that he's done using machine learning and AI, I was blown away by how novel his cause is and the stories he's reconstructing using machine learning and data science. And so that's what I'm really fascinated about. He's presented at the homecoming. And so we're going to get a deep dive into his research, his work. And so ladies and gentlemen, allow me to welcome Dr. Abraham Liddell. Hi, Abraham. Hello, it's a pleasure to be here. Thank you so much for saying yes to coming to the pod and for your insightful presentation at the homecoming, you're a historian and you're also a data scientist, which is a powerful combination because of the use case I've seen in your research. Before we get into that, do you mind talking briefly about how you got into GIS, your background of history and where did that connection come from? My background is in history. I got my PhD. Um, from Vanderbilt in Latin American history with a focus on the Spanish Caribbean and West Africa specifically, more so a broader focus on Atlantic history. I was motivated to study the movement of free and enslaved Africans and Afro descended populations and their social networks. A big component of my research was centered around being able to track individuals. As that evolved, I became involved in the humanities and data science projects while I was still a graduate student, and I began to see a number of ways that I could employ computational methods to get better insights into how people live their lives and to track things like social relationships as those. That's formed over time, kind of manifested into a postdoc in data science at Columbia University, which then turned into a research position here at City College, where I work as a associate research data scientist, helping to essentially apply the tools and methods that I had been Developing as a graduate student and postdoc to studying the early population of 16th century Las Manolas. It really began primarily as an interest in how can I find individuals records, particularly free or enslaved Africans. And not only how can I find them, but then it evolved into a more complex question of how can I track them and make some sense of their world and the kind of relationships that they formed. In addition to that, how can I. construct ways to try and understand how those relationships started to inform their lived realities. So it started off with the simple premise of being able to find individuals and then being able to track them and then trying to figure out ways to more accurately reconstruct the world that they lived in via the spatial and social connections that framed their universe. Thank you. That's really powerful. And what really stands out with this use case is, is that how novel it is, right? There's no limitation to what you can't do with GIS. And from an insightful point of view, when it comes to tracking and understanding their lives and how they're connected, what are some of the insights you've discovered with these stories? Yeah, so one fundamental part was thinking through the significance of social connections, and how vast their networks were, even though an enslaved person might be separated from their family by a large distance. If you had a connection with someone who was part of a very mobile category, like a Sanger, that opened the door for a large reservoir of social resources, because that individual could connect into a larger diaspora of people across that are connected to all these other hubs. In this Atlantic economy, you have a geospatial component. An individual might be seemingly isolated in one location but if they have a large enough social web, that links them to a number of other geographical places that can allow information, as well as other additional resources like perhaps Or in one case, someone's literal ability to get free after being illegally re enslaved. By blending this concept of thinking through geospatial analysis, but also thinking from a social perspective, I was able to make sense of how someone's world could be upended, but then simultaneously they could find their own freedom by relying on these very extensive social networks that link them to all of these other places. Thank you. One of the things you mentioned during your homecoming presentation was for this kind of research, How important it is for it to get out of the world of scholars and researchers and just get it out there. And so as you're speaking, I'm wondering when it comes to the impact, what does that look like moving it outside of academia? And how would you want to see that evolve outside of that world? Yeah, in my current role, a big part of what we have planned is to rely on the research institute, which is more part of the Dominican Studies Institute. You're at City College. They are very much community focus. There is a large Dominican population here in Harlem and in Washington Heights. One of the components that we're very excited about is taking digitized records of historical documents, transcribing those and extracting individuals from these records in 16th century. There's a significant amount of data that is. Vital to understanding the early Korean from an economic standpoint as well as a social and cultural one that project of that sort make it easier for the public to interact with that information stored in these records was in order to read them. For example, you have to have a particular set of skills both in. Spanish, Portuguese language, but also the capacity to read the handwriting, which can be a barrier to scholars who may not be trained in 16th century hands, for example, or if you're a layperson who is just very interested in studying your own historical past or the past of the Caribbean or Bronx, it's a limiting factor. But transforming that information into things like where all the individuals have been distracted and various kinds of information become available to the public, it makes it easier for interested people to ask questions to analyze data on their own to. Raise questions about geographic information, which is also stories, raising questions about people's mobility, where they were coming from, where they were going, being able to map that quite literally using GIS tools makes it so that a broader public can make sense of this data. It makes historical research. Significantly more accessible because it goes through a process of transformation. By transforming it, it changes it, but it also makes it so that it's digestible and accessible and inform a wider audience and a larger number of people. Always, of course, be people who strictly want to read the documents. You start to get a sense of the depth of their connections, and you get a broader sense of where they are. So that gives you the capacity to not only think about people moving and information, but also raw physical things. And of course captives, right? Captives are entering the colony, being able to map more accurately where they're coming from, vital part of historical research. And it's a powerful tool to see individuals. The Spanish had a habit of recording where individuals came from via their surnames. They would ask them where they came from. Some of it was because of the work that they required them at via these merchant diasporas. But sometimes you're able to reconstruct. Some of these ethnic communities that arrive in places from West Africa. You might see someone with the last name like Fulano, Rand, Mandingo, or Congo. A bit less in the early period, but a bit later you start to see this increase of people from those regions. And so you're able to create. Maps that reflect that, right? There's various ways of interacting with that kind of information. The big part of the current project is to go through this process of transformation but also make it so individuals can visually see, by using GIS tools, spatially where people are, where they move, who and what places they have connections to. It makes it more accessible and it adds depth to our understanding of a place. When you add these other dimensions. And even as you're talking, I'm just thinking it would not be possible to make all those connections without the work that you're doing. If I'm looking at a hard copy manual, not able to tell these stories, make these connections and track interest where it is coming from in the flow. Oh my God, that would be such a powerful link map. I'm obsessed with link maps and knowing directional flow and where things are going and where they're coming from. And so that creates a lot of context into. It's such a rich flow of information you're creating to make it make sense. I'm also really connected with this story in the sense that there's a lot of identity that comes with knowing historical pasts and what that sort of would evolve to in future. I'm, I'm relating what you're doing to remote sensing, which is a lot of pre processing and How much of a headache that can be. And it's literally me getting data sets from Landsat and it's already there, right? But you're going through hard copy records or historical records. So what would you say are the most underappreciated aspects of your work? I think anyone who works with data would say that it's really all in the processing phase. Because you're dealing with handwritten images, that requires a particular type of approach from a machine learning perspective. So essentially, what it is dealing with images with handwritten text recognition, which is a branch of computer vision. So, thinking paleographically, there are different hands that vary. And some hands are going to be a bit better. Sometimes they are recopied at a later date, so they're a bit more neat. And they're a bit more structured. Other times they were written by different people for different purposes. And so the writing is not very legible. There are other ways of processing face deals with dealing with some of these problems. Things like through, if people are all writing with ink and sometimes the ink bleeds from page to the other. And so you have this overlap that's not necessarily easy to remove depending on what kind of images you're working with. Also, sometimes just the image quality themselves, right? So if you, you have some images that are at the time when they were digitized or put on microfilm. It had a really bright light centered on the image, which causes some problems. Sometimes it makes some of the text illegible, even with the human eye, so that it's figuring ways around essentially processing the images so that they can be used for machine learning purposes, because the machine learning component is the sexist part, right? And that's the part where you're training the model, you're getting outputs, you are making iterative process, but. The data processing part, the image processing part is a fundamental part. And then in addition to that, the other part is dealing with the language itself. So obviously it's in Spanish. But given the time in which it's written for additional analysis, like text analysis and using batch language processing on the transcriptions, you do have to have some idea about the language itself. And removing artifacts from the images, depending on where you're sourcing images, maybe there's some damage to the images themselves, or to the document itself, that can obscure certain words or make them completely illegible. So these are things that you have to work around. And figure out how to solve. So, yeah, it's, it's that I think is an underappreciated aspect. Interesting. And because now when you're talking about the work that goes into that, I'm also thinking, how do you create a balance? Because these are human records you're handling when it comes to the ethics of balancing the physical records and knowing that these are actual. Human records and how do you handle that in terms of ensuring the information is not lost or the pages are not lost. Thankfully, the records we're using have already been digitized and we have an agreement with the Spanish archive, but they have already been digitized and have already been managed by the archivist there. But of course, even though we're using images that have been digitized, some of which are on microfilm, they do still have some aspects of them that make them difficult to work with how they were. Digitize the process they use. Sometimes some of them have really bad lighting. So that creates some really bad dark spots and some overly bright spots. So there's, and there are some of them do have artifacts on some of the pages themselves that have to be removed. But as far as physical handling of the records themselves, thankfully, that's already been done by the archivist. We solved. Okay. And I'm very fascinated by emergent trends and how they can shape how we look at our workflows. And yeah, we're definitely in this whole. Machine learning, AGI, this whole new world, right? And so I'm curious about, are there emerging trends you're excited about and how you see that shaping your work? I'm super excited about a lot of these tools. I think that the generative AI trend has a lot of use cases, especially believe it or not, in removing certain kinds of artifacts from machines. Using generative AI to reproduce the images without the artifact. It's actually a very powerful tool, something that we actually are employing here. And of course there's, when we talk about accessibility, I think an increasing use and domain specific large language models to provide insight into topics. It makes it more accessible than just a standard user interface. Being able to prompt a large language model to ask questions about a specific area or specific domain is a growing area and we're getting a lot better at. Figuring out ways to have domain specific linguist models, be able to answer more complex questions, and I think not so much necessarily for researchers immediately, but for general public use of people who are fascinated in a particular area that say, in your case, environmental sciences, having a large linguist model that is capable of handling a variety of questions reasonably well. Could make it so that people are able to learn or at least get insight into how the field works and learn new things and make it more accessible in ways they otherwise may not. I think that's a great thing to be excited about. You have this growing field of computational science relies on things like machine learning. And so I think we're starting to see this kind of what I think is a really cool transformation by using these new tools. And of course, now we're seeing it at the, my project. And hopefully more and more rising using machine learning and deep learning to not only transform historical data, but to answer questions about it to generate new areas of inquiry. And, of course, climate science, environmental science, and also geospatial technology, right? These things are going to also be transformed by things like. Machine learning and deep learning, like the internet of things is an exciting adventurous, right? The other bit of it would be the ethics component of it. I think you briefly touched on that at home coming during the presentation, the ethical considerations that come with handling historical data and also maintaining the human side of it. That's a good question. I think that one of the things that we have to consider is that the records. Particularly if you're dealing with records that are related to incitement, are dehumanizing records. People don't appear in them in ways that, in which they are seen as human. They appear in ways, so the records themselves aren't human biased. The goal is not to portray incited Africans or indigenous people in humane or meaningful ways. They're structured in a way in which. You have a particular perspective that is colonists and your cybers themselves who are interested in collecting certain kinds of information, right? What that means is that you have records that are inherently biased So one thing that you have to be considerate of I think is when you're dealing with this kind of information Is you have to try to make? The context around it, a big part of how you discuss it. What you don't want to do is to decontextualize the data that you have. That's a big part of trying to explain why people are living the lives that they are living, is the context. Data itself is an important component, but you always want to make sure that it is contextualized. Because without that context, It can, at times, reduce individuals to, as my advisor used to say, like being counted. You have this collection of people, they're just a number. You don't want that. The one important component, too, is that with the advent of machine learning and these flying computational methods to studying historical past, you can now look at things in aggregate in ways that we couldn't previously. And so, that in and of itself is actually a humanizing component. Being able to see individuals. In a multitude of ways, and also with a greater degree of accuracy pinpoint, essentially imagine kind of plucking one individual from a place in time and be able to see a lot more about them. So you're able to get both a micro and macro scale type of analysis that would be a bit more difficult to do previously. But I think what needs to be considered here is thinking about what these records are. Who's producing the records? Why are they producing the way in which they are? Adding this additional historical context to the information, because without that, the records themselves don't tell you what you might think they tell you. Context is a historian's greatest tool in explaining certain kinds of behavior, and I think that is a very important component. And it's also important that we consistently highlight individuals. For the sake of humanizing is again, we don't want to just see them just as a groups of people who exist in the abstract is imperative that we treat them as individuals. And I think we all have a responsibility to do that. And of course, there is always the question of making sure that whatever sources you're using to source these individuals and to source this data is acquired ethically. We do live in an area in which we do have access so much data. And so what that means is. Where, wherever you're sourcing information that you're using to train your models or that you're transforming to try and make available or to answer some kind of research question, you have to make sure that that you have permission from the people or places or countries or institutions that you're gathering the information. You have to make sure that you're always handing it in a way in which is in full accordance with your agreement. I think we live in an area that is, of course, very exciting. Artificial intelligence and its use cases has a large number of use cases. But what makes. Machine learning and deep learning so important is really the kind of quality and the kind of data that you have and making sure that you're acquiring that in a manner that is ethical and you have to take into consideration how you're getting it, where you're getting it from and making sure that you're using it in a way that is not harmed. And I think that has to be baked into the kind of work that you do. You're working with data of any type of any sort, especially as it pertains to people. There has to be some kind of process to make sure that it's collected ethically and it's done in a manner that's respectful to the subjects that appear in your data need to be treated as respectfully as possible. I think, given the fact that this is my domain, I feel a lot more comfortable in that respect. But I do think it's still something you have to do cognizant because the records themselves are human, especially because it deals with slavery and other various aspects of colonization. So do you want to consistently think of ways to highlight that these are people who are carrying these records are people worthy of respect and we should see their lives and struggles as meaningful and treat them with a level of respect. But even if that level of respect isn't always transparent in the records about. So there is still a process of transformation of thinking about. How are we consistently respectful of the people who are paying these records? How do we humanize them? Even in the records themselves, in which they are not humanized, are literally thought of sometimes as cattle or objects. So the ethics are something that you have to constantly, in a way it's an inner process. You're consistently thinking about how can we maintain a respectful approach to talking about the people who are paying these records. Because these questions, despite them being part of the past, also still maintain a level of present or contemporary importance. And so that is something that I think is unique to having a project that is centered around the historical past, is that There are echoes into the present. Right? There isn't, you're never fully removed from slavery, transatlantic slavery, the colonization of the Americas. These are still subject matters that are sensitive and still have modern repercussions. And so it does require a level of care and thinking about humanized individuals that appear there, seeing them as people and trying to make sure that whoever. Interacts with the records themselves, understands the context around them, and provides some context of the world in which they, why maybe they behave a certain way, or why maybe they experienced some kind of phenomenon. You want to make sure that all of that is transparent to anyone who might use them. I would tag that as a masterclass for data and ethics and how we should handle data, because you've talked about so many contexts and you've brought the points home on what the ethics side of it looks like when handling records and the contextualization and being able to humanize that as well. And that was very insightful. So I'm going to go back to the theme. And this year we're talking about art, Afrofuturism and geography. And so based on how you've painted out the picture of what ethics looks like and the handling of these records, I'm curious to know how your perfect utopia would look when it comes to these themes, Afrofuturism, Art and Geography. A perfect world? That's a big question. I've always liked the concept of Afrofuturism. I think envisioning a world in which Black people globally are safe, secure, and free of many of the oppressive forces is reappealing. The global phenomenon is something like Black Panther, for example. Despite its very Americanized concept of a African nation, which has many of the problems that an American viewer would have, especially because it was constructed by white men, which sometimes gets left out, that Wakanda, the concept of it, Black Panther is entirely constructed by white men in the west. The concept of society in which black people are living in the future, in which our lives are not strictly based around suffering and oppression, is something that is imagination, I should say, is a critical point, right? We want to imagine worlds in which we are living, Safely and in community with each other. I think a perfect world for me is hard to immediately come up with off the top of my head. But I do have some things that I think are central to that kind of a better world, which would be one community. I think particularly in the West, our concept of community is significantly different than other places. I would say that community here is not very strong or sensitive among Western nations. I think it's strong among the Black community. But. The concept of living in community with each other and being responsible to other people is a big component of living in a world. I don't know if you're familiar with the concept of punk at all, but it is, I think, a very fascinating. Solid Punk is a fantasy where it is a distinctly community based world in which people are able to live in harmony with the environment. I would say it's also distinctly pro human in the sense that perfect worlds would be centered around this idea of how do we treat all the aspects of our humanity and how do we take care of ourselves and not just see ourselves as on producing something labor. Labor is a part of our lives, cannot take up the entire spectrum of our lives. It is a small part. And I also think a future world looks like us using technology less as a means of just producing more and making more money, more in a way of legitimately just making life easier, while also at the same time, helping us to manage the environment in a way that's not destructive or distracting. So I think, uh. A future world is going to be rooted in us reconsidering how we think of our relationship to each other and our responsibility to each other, but also one in which we use technology with our primary focus is not necessarily strictly for extractivist reasons or approaches, but also for how we use technology to make our lives better, to make things more equitable, to make things more accessible. And I think that is a shift strictly in its. I think technology currently, while that is certainly a component of it, is through the lens of acquiring capital and making a few people very wealthy, right? There has been a enormous growth in wealth in the tech billionaires currently. Most wealth, I think, doubled in the past three years, maybe? I think in a perfect world, there wouldn't be that level of gross inequality, and there would be a greater sense of connectivity and responsibility to influence. I profoundly believe that we can solve all of our problems. It's really a matter of priority. In a perfect world, I think we would more prioritize these kind of social and environmental problems, right? So it's not really a matter of can we do it, it's a matter of whether or not we have the will to do it. I think a perfect world we have shifted to a more social and cultural awareness of each other, and that shift will then lead to changes in how we approach the use of technology, and changes in the way in which we live in the environment. And how we live socially and culturally. So in a perfect world, we move away from thinking of extracting constant growth and more towards a sustainability, slower paced, rooted around caring for each other, the environment, and figuring out ways we can use all of our resources to solve these very large and complex problems in a way that is humanizing. To me, I think that is one of the more perfect world. I like how profound that was, and I also like that it caught you off guard, and so I'm gonna cut you off guard again with another one. What was in your 2024 bingo card? Yeah, lots of stuff was in my bingo card. Yeah, a lot of things, I would say. Yeah, so many things have happened in the past year, in the current time, so I think it's something that is actually really fascinating. Okay, staying in the spirit of deep learning and machine learning, I did not have, I don't know if any of you had a chance to play with some of these models like Sora made by OpenAI. I did not have them being that good on my bingo card. It's a little concerning for ethics reasons. So I don't know if you guys know what Sora is. It's a text to video model, so it's a generative model that takes A text prompt and it's able to generate videos that are really quite good and convincing enough. Where they were like two years ago was nowhere near how good it is now. Like the rate at which they're improving is a bit concerning. If you guys have elderly parents or grandparents or the youth are easily deceived by fake videos, things that are simply not real, it makes that a lot easier. So I did not anticipate it being that good. That quickly considering where they were, it feels like we are in the midst of multiple shifts. And so it's really hard to predict outcomes. I didn't have that. These major shifts on my bingo card. I did not have. Yeah. So many things have happened in the last year. Here we go. Oh my God. I can't wait to see what this means from a technical perspective and what that would look like for us. So yeah. I'm glad I chipped into that. I'm so excited about that topic. So I think we're going to, we'll wrap it up. Thank you so much, Abraham, for having this conversation, coming to the board. I think for our audience, you've had a view of what it looks like when tech and history converge, and we've seen powerful use cases that are historical, but resonate with us, Currently in but we identify with our stories tracking where we've come from, our roots and where tech sits within that. Once again, thank you so much, Dr. Liddell for coming forward and we definitely look forward to hearing more from you and following you as well or this and following your work as well. So yeah, thank you. so much for having me. If you enjoyed this episode and want to learn more about Northstar of GIS, check us out on Facebook, Instagram, and YouTube at GIS Northstar. We want to thank our sponsors of the 2024 Homecoming event, our institutional partner ReGrid, and our sponsors New Light Technology, Afrotech, and Black at Work. We'd like to thank our keynotes, Tara Roberts, Linda Harris, Dr. Paulette Hines Brown, and Vernice Miller Travis. We'd like to thank Howard University and the staff at the Interdisciplinary Building and Photography by Imagery by Chioma. We also want to thank our guests for trusting us with their stories. Tara, Linda, Paulette, Christian, Abraham, Jason, Vernice, Stella, Beye, Karen. Nikki, George, Frank, Labdi, Toussaint, Victoria, and the HBCU Environmental Justice Technical Team. And finally, thank you to the North Star team and our wonderful volunteers. We are your hosts of the Season 2 of the North Staggers Podcast, which is based on the 2024 Homecoming Conference event. Thanks for listening to the North Star Gaze, intimate stories from geoluminaries. If you're inspired to advance racial justice in geofields, please share this podcast with other listeners in your community. The intro and outro are produced by Organized Sound Productions with original music created by Kid Bodega. The North Star Gaze is produced in large part by donations and sponsorship. To learn more about North Star GIS, Check us out at north star of gis.org and on Facebook or Instagram at GIS North Star. If you'd like to support this podcast and North Star of gis, consider donating at North star of gis.org/donate or to sponsor this podcast, email podcast at north star of gis.org. You've been listening to the North Star Gaze.