November 20, 2026

00:48:41

Finding Truth in Data Visualization

Finding Truth in Data Visualization
UVA Data Points
Finding Truth in Data Visualization

Nov 20 2026 | 00:48:41

/

Show Notes

Can a beautiful chart also be misleading? And where do we draw the line between persuasion and truth?

From color and typography to the way information is arranged, design choices shape how we see, understand, and interpret data. But how do we balance visual appeal with scientific accuracy?

In this conversation we have two experts who approach these questions from different perspectives: Alberto Cairo, a leading voice in data visualization, author, and creator of the Open Visualization Academy; and Alex Gates, assistant professor at the University of Virginia School of Data Science.

Together, we'll explore the intersection of data, design, and storytelling — and what scientists and artists can learn from each other.

Chapters

  • (00:00:01) - Introduction and Guests
  • (00:01:11) - Alberto Cairo's Origin Story
  • (00:03:40) - From Infographics to Data Visualization
  • (00:05:34) - Visualization as a Way of Understanding
  • (00:07:54) - Visualization as an Iterative Process
  • (00:09:51) - AI and the Future of Visualization Craft
  • (00:12:19) - The Reader's Responsibility and Hurricane Maps
  • (00:17:13) - Teaching Design and Reading Together
  • (00:19:13) - Is Data Visualization Art?
  • (00:20:45) - Aesthetics vs Understanding
  • (00:22:56) - What Scientists and Artists Get Wrong
  • (00:25:38) - Math, Statistics, and Collaboration
  • (00:29:30) - Journalism, Data Science, and Identity
  • (00:34:10) - The Open Visualization Academy Origin
  • (00:39:20) - Evolving Teaching Over 15 Years
  • (00:41:12) - Generative AI's Risks and Cognitive Capture
  • (00:43:40) - Taking Ownership of the Tools
  • (00:48:12) - Closing Remarks
View Full Transcript

Episode Transcript

[00:00:01] Margaux: Welcome to UVA Data Points. I'm your host, Margaux Jacks. Can a beautiful chart also be misleading? And where do we draw the line between persuasion and truth? From color and typography to the way information is arranged? Design choices shape how we see, understand, and interpret data. But how do we balance visual appeal with scientific accuracy? Joining me today are two experts who approach these questions from. From different perspectives. Alberto Cairo, a leading voice in data visualization, author and creator of the Open Visualization Academy, and Alex Gates, assistant professor at the University of Virginia School of Data Science. Together, we'll explore the intersection of data design and storytelling and what scientists and artists can learn from each other. [00:00:50] Alex: Well, thank you so much, Alberto, for joining us today. Given that we're here tonight celebrating Data is Art, I can think of few people better positioned to help us think about what happens when data, evidence, communication, and creativity all come together. So I think the best way to start is at the beginning. And I'd love to hear a little bit about your origin story. You've said that you came into visualization almost by happenstance, [00:01:14] Alberto: not almost completely by happenstance. [00:01:17] Alex: So can you take us back to the beginning? What were you actually trying to do? And how did visualization enter the picture? [00:01:22] Alberto: Oh, man. Well, first of all, thank you for having me. It's a pleasure to be here. Great honor. So how did I begin? So, I mean, as you know, right now I am a professor of visualization and information design in the School of Communication at the University of Miami. But that is after 30 years of designing graphics, mainly for newspapers and magazines, in the context of journalism. But that is not why I studied journalism. So I have a BA in journalism, but I didn't study journalism because I wanted to become an information designer, a communication designer. I wanted to do radio. And I actually, when I was in my 20s, I did an internship at the sort of like the equivalent of npr, National Public Radio in Spain, reading the news in the morning, which is a job that I still enjoy. As you know, I have my own podcast, the Data Journalism Podcast, Product placement here a little bit with my friends Simon Rogers and Scott Klein. In any case, I always enjoyed audio as a medium. But the thing is that my grades during my BA Were not that high because I didn't pay a lot of attention to classes. I was doing a lot of other things on the side, writing for the university newspaper, having my own radio program. I didn't pay a lot of attention to classes. So my grades were relatively low. And in my last year in journalism, I couldn't get an internship in radio. And there was this professor at the university who received a request from the graphics department of the local newspaper where I was born. The newspaper is La VOZ de Galicia. I was born in Galicia in northern Spain. And this professor received these requests saying, do you have any student who is studying journalism and who has sort of like an eye for design, who can draw, who can sketch? And I happen to know how to sketch. I like to draw. And this professor knew this, and she recommended me to become part of the be the intern at the graphics department of this newspaper. And this happened in 1997. And it all, well, downhill from there, essentially. That's how it happened, [00:03:35] Alex: yeah. And so then how did you start incorporating data into those visualizations? Is it graphic design? That's another story. Visualization. Yes. [00:03:43] Alberto: Yes. Because initially, between 1997 and I would say, 2009 or 2010, most of work was more on the information design space, or infographics space in the news. Infographics used to mean pictorial representations of information. So, for example, if there is an accident or whatever, you create sort of like an animation or illustration and explain how it happened. If NASA sent a probe to Mars or whatever, you created an infographic explaining sort of like the structure of the machine. You did a cutaway. So I did a lot of 3D modeling and visual explanations. That's what I was doing. But what happened was that around 2008, 2009, I told myself, well, you know what? The maps that I have been designing so far, when I look at maps designed by actual cartographers and professionals, my maps look like crap. I said, I actually need to learn a little bit more. Learn a little bit more about cartography. So I started educating myself on cartography. I got a whole shelf of books about cartography, and that led to Data cartography, how to depict data through maps. I found that absolutely fascinating. And the readings that I was doing about cartography led me to do more readings about data science, statistics, and eventually data visualization. That's how I transitioned again in the early 2010s between pictorial representation and data visualization. That's how it happened again, by happenstance, it was like, all right, [00:05:17] Alex: I love that trajectory. So you've started with the storytelling, the pure audio aspects of it. Then you moved into information and trying to understand something, and finally you integrate it in the data. Yeah, [00:05:27] Alberto: that's a really good summary. I've never thought about it that way, but you're completely right. That's how it happened. [00:05:32] Alex: Nowadays, our students come at it the other way. Right. We always start with the data and then we dive into trying to understand what's happening and move further that way. Yeah. So I mean you've often described the visualization is not just a way of presenting what we know, but it's a way of understanding and seeing something. Could you kind of explain a little bit more about what you mean by that? [00:05:51] Alberto: Yeah, I mean, I see visualization in many ways in a very similar way, way that I see writing, reading, actually reading and writing, both of them. To me, reading and writing are not just about acquiring or conveying information, they are ways of reasoning about that information. So by reading about, when I read a book, for example, the action of reading is not passive. I am creating a dialogue with the author, I'm taking notes, I am sort of like debating with myself. And when I write it is the same thing. I struggling with myself, I'm having a conversation with myself about the information that I'm writing about. And to me, visualization is very similar. Through the act of visualizing information, I get to learn the information at a deeper level, a more structural level, so to speak. So for me, visualization has sort of like a very strong exploratory component. Again, this is not news. I mean, we have the old classic exploratory data analysis by John Tookey who spoke about these. Right. He said, Tookey said in the 60s, if you want to understand your data set, always visualize your data. Now this sounds like a no brainer today because visualization is everywhere. But it was groundbreaking in the 60s and the 70s because statisticians at the time, what we could call today data scientists, although the term had not been invented yet, I would say were very wary about graphics. Statisticians in the 60s and 70s and before thought that visualizations were just sort of like toys, they were like jokes for the public and they didn't take these tools. Visualization as a serious component of Inquire. Right. And that changed. It changed through Tukey. I also see visualization obviously as an extraordinary tool for communication, which is what I specialize on. So I don't just write or speak, I also visualize and I combine everything in sort of like we could call them multimedia experiences. Right. You convey the information through the words, but you also convey information through the visuals. [00:07:53] Alex: Yeah. And I think what really resonates out of that is that you also are talking about visualization more as a process than a one-off stack. Right. There's an engagement of that process of actually trying to understand what's happening in the data, repetitively improving your visualization. It's [00:08:08] Alberto: an iterative process. I mean, that's something that actually surprises my students. So many of these data science students come to my class, what I would say an engineering mind. They expect, and I'm not trying to sort of downplay or. Engineering is obviously a very serious series of fields and knowledge domains. But they come to my class, many of them expecting that I'm going to teach them rules. Do these, don't do that. Use this graphic in this context. Use this other graphic in this other context. They expect that those rules are set in stone. The first thing that I try to do in semester is to disabuse them from that notion. It's like I said, this is a process. As you say again, it's like writing. Are there rules in writing? Well, beyond knowing the grammar and the syntax and the vocabulary of your language, what you do with your language is largely a process, it's largely a reasoning process. And what I try to instill in them is this sense of reasoning about the different factors that contribute to a visualization being effective or not being effective, which are very context dependent, are depending on the nature of the data that you represent, depends on the communicative intent that you have. It depends on constraints such as the size of the visualization, all that need to be weighted in in order to produce the visualization. So it takes obviously a long time to produce a good visualization. It's not just about loading the data into a software tool and clicking on a button or prompting an AI to create a visualization for you. If you are doing it that way, you're not doing it the right way. I would say [00:09:51] Alex: so. I mean, that hits on a really big crisis right now in academic scholarship, which is that the writing process through which we discover the scientific arguments is now being replaced with generative AI models. And similarly the visualization process is now being replaced with generative AI models. So kind of do you think that there's still a place for visualization in the world of AI? And why and how do we still need to understand that visualization design process when you can just go and generate a decent image with the click of a button? [00:10:27] Alberto: Well, I mean, you say to yourself, are you satisfied with a decent image or do you want to produce a great image? That's the thing, that's the difference. It's like, I think that we should aim for excellence, not mediocrity. And what AI produces on its own is usually mediocre. It's good enough, but it is not great. In order to make it great, you still need the intervention of a person. I mean, AIs, particularly the frontier models, are getting better. So now you prompt them and they create a pretty decent image. But still they are not aware of the context, they are not aware of the audience. They are not aware, again, of all these different factors that you need to weigh in. That doesn't mean that you cannot produce a good visualization with a generative AI tool. You can, but you need to use the AI in a very similar way that you will use a programming language or a click and point tool. You need to guide the process. Again, it's an iterative process. So if you want to create a visualization the same way, the only difference is that instead of using a graphical user interface to produce the visualization, you use a natural language interface to produce the visualization. But to me, what underlines all that process is the knowledge that the person has, which you cannot offload to a computer, because if you do that, you're going to lose that knowledge. You need to develop the knowledge before you use the tool or while you're using the tool. You need to understand, for example, why the tool does certain things in a certain way. Why are we putting the baseline at zero? Or why are not we putting the baseline at zero? Because some graphics need a zero baseline, some graphics don't need a zero baseline. So how do you make that? How do you reason through all those types of decisions? The computer can do it for you, but then essentially you are putting those decisions in the hands, so to speak, because AI doesn't have hands of a black box system. And I don't think that that is healthy, obviously. [00:12:19] Alex: Yeah, I completely agree with you on that process. But I think one of the things you mentioned is that it's not just about the specific story that you're trying to tell, but it's also about understanding who you're trying to tell that story to. The role of the audience and the viewer in trying to understand the visualization and as a key component of the visualization process. So can you talk a little bit about that? Like, how much responsibility does the viewer actually have to be visually literate and to understand these different visualizations. And is it the designer's job to prevent all the misunderstanding, or does the reviewer actually have agency and responsibility in understanding what they're seeing? [00:12:58] Alberto: I would like to thank you for giving the answer, because I do think that readers have a responsibility. Not all misunderstandings of visualization are the product of bad design decisions. The design process is actually an interaction between the creator of the visualization and the readers, the audience of the visualization. This is something that I covered in my third book, How Charts Lie, which is sort of like a joke. The title is a joke because charts cannot lie if somebody doesn't lie with them. And the book is filled with examples of charts that don't lie, but they still mislead because they are misinterpreted by people who should have known better or who are not prepared to read that type of chart. Right. The challenge here is that for many years I think that the visualization literature, I'm not talking about the present, I'm talking about the past, focus too much on the designer side, trying to teach the designers best practices rules. Although I don't believe in rules, it put a lot of the wake of this decision making process on the designer side. I think that that is wise. The designer has a huge responsibility to ensure that the visualization will be as easy to understand, as deep and as true to the as possible. But that will not guarantee that visualization is going to be successful because you will never know whether the readers that are going to find the visualization will have the same knowledge that you have about the topic that is being visualized. The perfect example I have for that which I describe in the book also, there's a whole chapter about this. It's an area that I have been obsessed with for many years is hurricane mapping. Obviously I live in Miami, so I get to see the national hurricanes right [00:14:46] Alex: in the center of every map that's been produced. [00:14:49] Alberto: Yeah, yeah, yeah. The thing, the cone of uncertainty. Right. Everybody who sees that map for the first time misunderstands that map. Just again, it's difficult to talk about visualization using audio, but try to envision, this is for the listeners. Envision a map that represent that has a cone in the middle of increasing width. What does that cone represent if it is representing a hurricane? Well, obviously it's representing the path of the hurricane. But what does the cone represent? Well, many people interpret the cone as an area under threat. If you are inside of the cone, you're in danger. If you're outside of the cone, you're not in danger. And we have observed these. I'm part of a research group at the University of Miami with several colleagues of mine. We're studying how people misinterpret the map and what can be done about it. And again, that is not what the cone is representing. The cone is a representation of uncertainty. It's Like a probability distribution. So essentially the cone is the area where it is more likely. Likely that the center of the storm, center of the storm only, that's really important, may be in the following five days. But the center of the storm may lie outside the boundaries of the cone. That's a much less probable future, but it can still be outside the boundaries of the cone. So in order to understand how to read the cone, you need to be taught how to read the cone. The problem with these types, with all types of visualization is that there's this misconception related to the old saying, a picture is worth a thousand words. That's absolutely not true. A picture is worth a thousand words if you know how to read that picture. If you don't know how to read that picture, then it's worth nothing. And that's the problem. The problem with examples like these is that the designer creates the graphic with a particular mental schema in mind, a particular mental model of what the map is intent, or the graphic or the chart or the graph, whatever it is, is intending to represent. But the viewer may come to that map with a completely different mental model in mind. You are having a cone that represents a probability distribution. But then you have the reader comes to your map inferring, because they have seen previous maps in which a bounded area represents something that is geographically different to other areas. And that's the mental schema that they apply to the map. And that's where misunderstanding happens, in the mismatch between mental schemas, between mental maps. I'm sort of like getting carried away. Feel free to interrupt. [00:17:24] Alex: No, no, but I think it really speaks to the importance of modern data science education right now. Right. Where traditionally everybody would think of a data visualization course as being a course where you learn how to make your own visualizations, [00:17:38] Alberto: which. [00:17:39] Alex: Yeah, it's a core component of it. Right. But what you're speaking to is something much larger and that's building literacy on how to understand and navigate a world that's increasingly filled with data driven argumentation and typically presented visually. [00:17:53] Alberto: Correct. It's all about sense making. It's all about sense making. Visualization is part of sort of like sense making and sense communication, so to speak. Yes, [00:18:01] Alex: I love that. So maybe that kind of transitions. You're one of the most prolific educators that are out there in the data visualization space. How do you view that tension then between trying to teach a class how to design and teach a class how to understand design? [00:18:17] Alberto: I teach both. That's the spirit of both My classes at the University of Miami and also my latest project, the Open Visualization Academy, which is a free and open source platform for visualization education with tons of free courses. I say very explicitly at the beginning of both my classes and the Open Visualization Academy courses that learning how to design visualizations goes hand in hand with learning how to read visualizations. So you cannot be a good visualization designer if you're not a good reader of visualizations. And I would say that also people who learn really well to learn how to read visualizations really well, the only next step that they need to take to become a visualization designer is to learn some tools to create those visualizations. But you cannot have the design without the reading component. They are intrinsically connected. [00:19:13] Alex: So I'm going to shift topics just a little bit. And given the theme of tonight's event, Data is Art, I wanted to ask kind of a dangerous question. Is all data visualization art? [00:19:25] Alberto: It can be. I mean, I don't have. I'm not an expert in the philosophy of art, obviously, but I have a very pragmatic, philosophically speaking, pragmatist approach to many things, and art is one of them. Anything can become art, right? Depending on the types of values and purposes that you imbue in objects. So visualization can be artistic the same way that, I don't know, any object can become art depending on the context that you put that object into, and depending on the approach that you have to that object, depending on the messages that you want to convey with that object, depending on many other factors that have to do with the intent of the design. At a much lower level of the discussion, any visualization has also artistic components in the common sense of the word. You need to care about typography, color, composition. Those are sort of like artistic interests as well. So even if you are designing a visualization just for pure analytical purposes or business communication purposes, there is still an artistic component to your work, that's for sure. And then there are people who use visualization as a medium to express themselves or to express ideas in more creative ways. They create art with visualization data art, so to speak. So, sure, I mean, I see those worlds as intrinsically intertwined, [00:20:49] Alex: but one of the complications to that is that there's sometimes a tension between the understanding of the data that's underneath and the beauty or the emotional evocation of the. I [00:21:00] Alberto: don't see [00:21:00] Alex: them [00:21:00] Alberto: as sort of like contradicting each other. I would say that again, going back to this discussion about that, I see visualization As a reasoning process, one of the things that we need to weigh in this decision-making process, reasoning process, is again the purpose. Some visualizations need to be designed for quick understanding or for exploration or for. You need to think about. And sometimes visualizations can be designed just for fun, and that's perfectly fine. But you need to define the purpose and then you need to be true to that purpose. I don't see that there's a huge tension between making a graphic understandable and making a graphic beautiful. I think that it is possible to do both. And following, for example, Donald Norman, the author of the famous The Design of Everyday Things, he also has another book titled Emotional Design, in which he says things that look good tend to also work better because for some reason things that are beautiful, if they also are, well, structurally designed, they tend to communicate better, they tend to work better. Maybe it's because they put us in a good mood and maybe it's because we get attracted to beauty in some sense. So I don't see a contradiction. It is true though, again, going back to your question that again, it's extremely relevant to think about what your intent is and then prioritize that. If your purpose is to communicate clearly or your purpose is that people will be able to extract very precise measurements or make very precise comparisons, you need to prioritize. Obviously, the more sort of like a structural alignment of the visualization, not the sort of like the surface level, sort of like aesthetics of the visualization. But that doesn't mean that you need to sort of like downplay the importance of the aesthetic components of the visualization, because they do have a role. [00:22:56] Alex: Another thing that this event deliberately does is it puts data scientists and artists right in the same room, which is something that doesn't happen nearly often enough. So I'd love to ask you, like, what is something that scientists tend to get wrong about communication and visualization, that artists and designers understand? [00:23:13] Alberto: Well, the value of the emotional component of the visualization and the value again of making things, making things look nice, look beautiful. To me, that's again, when I speak about these things out loud, they sound silly because obviously you need to care about beauty. I mean, one of our roles is to put more beauty in the world, to make the world more beautiful, to make the world more welcoming to people in general. And when I speak out loud about these many scientists, get it. I teach a lot of workshops for scientists about visualization. When I teach workshops, by the way, for scientists, I tend to emphasize all these aspects of visualization. Composition, hierarchy, beauty, beauty that doesn't interfere with understanding, typography, color, all that type of stuff. I make all these points. You need to care about these things. These things are not. There's like an expression that I loathe, icing on the cake, right? Aesthetics is not the icing on the cake. Aesthetics is intrinsic to effective communication. You do need to care about that. Now, many scientists, not many, but a few scientists, are a little bit wary about these. But I think that I'm very persuasive when I present this argument. They see it right. I get, for example, when I do a workshop for scientists, I ask them to send me charts that they have designed for their papers. This is also something that I do in my classes at the university. Give me your graphics and I will try to redesign them in a way that they will still be true to the data, presented data effectively. But I'm going to focus a little bit more on these more aesthetic components. And I put the before and after side by side. And I ask them honestly tell me which one of these you think work better, not look better, but work better. They choose the one that, besides presenting the information effectively, also looks better. Because again, looking better and working better, they go hand in hand on the other side. When I teach this work, these workshops to artists and designers, I focus on the other side of things. It's not just about creating cool pictures. It's not only about creating things that look extraordinarily good. You need to care about the structure, you need to care about the underlying. You need to be true to the data, so to speak. You need to present the data accurately. I also tell them visualization is not presenting what the data is, so to speak. It's more about presenting what the data means. And there's a lot of implications in this type of thinking. [00:25:41] Alex: Yeah. So then, what is the role of data science for the typical designer or artist, Someone coming not from the technical backgrounds? What would you recommend for them to understand about data science and the scientific process that could enhance their work? [00:25:54] Alberto: Well, they need to acquire, I think that designers could get. When I say designers, I also include any type of communicator who is afraid of math. I remember when I was in journalism, in my VA in journalism, some of my, my peers in journalism school used to joke, I am in journalism because I don't like math. I told them, not at the time, but later in my career. I sort of replied to myself in terms of saying, well, if you don't like math, you shouldn't be in journalism. You need to know Math, because without math, and as an extension of math, statistics, although statistics is not just math, it's also logic and reasoning. And so there's a humanistic side of statistics. I don't want to get into that. But in any case, that's intrinsic to communication. I mean, you need to, if you want to communicate information, first of all, you need to be able to interpret that information yourself. And math is an intrinsic part of that. So that's what artists, designers, communicators, journalists need to embrace more. By embracing math and statistics, I don't mean that you need to become a professional statistician. I mean, if you ask me right now how to calculate a standard deviation, probably I will need to sort of like do an effort of memory to remember the formula, Z scores and whatever. I don't remember it right, I will remember it eventually. And if you ask me for more advanced statistical concepts, probably I would ask you, Alex, just give me the formula for these. But I do have sort of like, I would say a working understanding of those concepts. What do they mean? What are they useful for? When should you use a median or an arithmetic mean or a mode to represent a data set in terms of measures of central tendency? When it is important to show the range and the shape of the data and not just that measure of central tendency and so on and so forth. I could go on and on. That doesn't require a deep understanding of the sort of like the mathematical structure or how you get to those calculations. It's more about this conceptual level of understanding that you can gain simply by reading or by popular science books or by talking to experts about all these fields. The other side of things is that. And you were talking about collaboration before. I think that this is incredibly important. Whenever I do make visualizations, I never make visualizations on my own. I always work in collaboration with other people because I lack domain specific knowledge about everything that I do. I'm a journalist, so I'm not an expert on anything. I just know how to translate things. And in order to be able to translate things, I need to understand them first. And in order to understand, I need to talk to you. I need to talk to a demographer, I need to talk to a political scientist, or to talk to physicists or whatever and ask them specific questions about their data. Okay, here's your data, but give me the elevator speech of what this data means. There's this wonderful phrase. I don't remember where I got these. Oh yes, I got these from an amazing book published in the 90s, statistics as principal arguments. That's the title. I don't remember the author. He was a professor of psychology at Yale University. So statistics as principal argument. The author of the book, in the introduction said that whenever I'm teaching statistics to students, they tend to focus too much on what is the right model, what is the right test that they need to apply. He said that comes later. That comes later. Begin with the following question. If you were to report about these data set in the newspaper, what would the headline be? I said, that's what I do. That's [00:29:28] Alex: journalism. [00:29:29] Alberto: That's journalism, right? [00:29:30] Alex: So statistics, you start from journalism. [00:29:33] Alberto: It's not that you need to start there, but it is an essential component of that. It's like at some point you need to ask yourself, what does all these mean to other people? To me and to other people as well? [00:29:45] Alex: Yeah, I think that that speaks to something that we deal with a lot in the school of data science and that's the interdisciplinarity naturally and inherently involved in data. Sc. We often joke that there's only two faculty members with a PhD in data science. Everybody else is from everywhere else. [00:29:58] Alberto: But journalism is the same way. What is a journalism degree worth? I don't know. Well, it's useful because it teaches you all these skills but at the end, if you want to get domain specific knowledge to collaborate with other people. [00:30:11] Alex: So one thing that I always struggle with is that differentiation from what my original training was and now this new world of data science science. So how have you found that transition and do you think of yourself now as a data scientist? Do you still believe that you're a journalist? Can they be one and the same thing? Or how does your identity, how has that evolved and how do you view yourself? [00:30:33] Alberto: Yeah, my identity has evolved quite a lot. I would not call myself. All right, so let me rephrase that. At the University of Miami, besides being a professor of journalism, still call myself a journalist and a designer. That's who I am, but that's part of my identity at the university. I'm also part of the leadership of the Frost Institute for Data Science and Computing. When they invited me to become part of that group, I said, I'm not a data scientist, what am I doing here? But the people who are there, who are actual real scientists, not only data scientists, but people coming from climate science, coming from environmental science, biology, whatever, biochemistry and so on forth. They told me, well, yes, you are doing data science. It's like whether you like technology or doing data. And I realized that the problem That I had is something that I always explained to my students. Remember what I told you before, that most of my students these days come from the data science program, but not only, they also come from other areas. I have engineers, statisticians, mathematicians, biologists coming to my classes to learn how to communicate better, how to design a better poster to bring to conferences. That's what I teach them, right? And some of them are a little bit wary or a little bit, sort of like they feel misplaced a little bit. What am I doing in a journalism class? And I tell them, well, the problem that you are having, and I'm going to disabute all these, you are going to learn all about this in just a couple of weeks. And they all understand the point. Join a journalism class. But that doesn't mean that you're going to end up working in a newspaper. Because journalism at its core is about trustworthy, true to the data, effective communication for the public good. That's what journalism is. It's about disseminating, collecting information, gathering information, filtering, interpreting and then conveying the information to benefit society. That's journalism. That's what journalism is. It's not a profession, it's a way of looking at the world. I have a friend, for example, Brian McNoldy at the university, who is also part of this research group about hurricane mapping. So Brian is a serious scientist. He does science, he does research, but he also has a blog about hurricanes, a web blog about hurricanes. And he writes regularly about hurricanes using common language, he doesn't use jargon. But his blog posts are really deep. I mean they're very informative, but they can be understood by a regular person and non expert, even if he goes really deeply into hurricane forecasting science and stuff. And to me, what Brian does, being a scientist and having his blog to communicate all these ideas, he's doing journalism. And I told him this is journalism, this is what a journalist should do, right? So I see data science in the same way. I learned to see data science in the same way. Am I a professional data scientist? No, I don't feel qualified to say that. But at the end it all comes with the practice. It's not the degree that you have, it's more about what you do. Again, this is more, more a pragmatist approach to how to see things in the world. We are what we do. How I would define going back to your original question, identity. So yeah, I still see myself as a communicator, a journalist. I see myself as a translator, a translator of Science. And more and more these days, I see myself more leaning towards a more philosophical approach to things. So eventually I would like to be able to call myself a philosopher or visualization. That's what I'm focusing on lately. [00:34:10] Alex: This sounds absolutely fascinating. And the more that you've talked about your educational philosophy and the types of courses that you teach, the more I really wish that I could take those courses. And what's so cool is that now I can. So you've made a major investment in the Open Visualization Academy. Could you tell us a little bit about that and why now? Why was this the time to launch? [00:34:30] Alberto: Yeah, well, that's another funny story, but happenstance so happened by happenstance. So at the university, I am an endowed professor. I am the Knight Chair in Infographics and Data Visualization. And as part of this position, I have my own personal budget that I can use to pay for travel to conferences, bring students with me to conferences, or invest in promoting other events or organizing events, conferences and stuff like that for different circumstances. One year, several years ago, I found myself in a situation in which my endowment had grown quite a lot. So it had generated plenty of money. Great position to be in. Yeah. But it's also a problem because I was focusing on other things and now I needed to spend that money on something right when I was reading and writing a lot. But yeah, it's a good place to be. So I said, what can I do with this money that will fulfill, that will connect with my identity? So I said, if you are a disseminator, if you're a translator, if you are an educator, what type of project can you create that will essentially disseminate this knowledge more at scale? Many Years ago, in 2012, I got into MOOCs massive open online courses. I actually created the first journalism massive open online course in the world in 2012. It was an introduction to data visualization and infographics for effective communication. I did this in collaboration with. With my friend Rosenthal Alves, who is the night chair at UT at the University of Texas. And it was sort of like a super improvised thing. So we did it with no budget. I recorded the video lectures at home. It was sort of like a. But it was a great experience. Even if we, again, we had no budget. It was sort of like this very scrappy improvised project. It sort of like exploded. I mean, we put it online and we got like 2,000 people signing in in the first. I don't remember first week or something to the. That we needed to expand the enrollment so to let more people in. And I realized and throughout the years, thanks to that first MOOC and several other MOOCs that I did in 2013, 14, 15 until 2016, I believe that I, or 18, I don't remember, I did MOOCs wherever I go in the world to teach, to give a conference, a talk or whatever. I usually get approached by people in tons of different countries, people who I have never met, who tell me I began my career thanks to that mooc. That MOOC was my first step towards learning about all this stuff. Today I'm a professional data scientist. I'm a professional data journalist. I'm a professional data journalist or data science or data visualization designer. I told myself when I found myself in this situation of having to make this investment, I said, well, I like that. That's what gives life meaning. If this is part of your identity, then this is what I should be doing at scale. And that's how the Open Visualization Academy, it's not a massive open online course, sort of like platform, it's a video course platform. So if you want to get a sense of what data visualization, not only data visualization, more broadly, information design. Because data visualization is part of information design. So if you want to get a sense of that, you can just go to Open Visualization Academy. There is an intro course by me, actually two intro courses. We just published another one yesterday, a couple of intro courses by me. And then there are specialized courses related to other topics. Cartography. There's going to be soon an introduction to statistics for designers that my friend Cameron Riopelle, who is the data librarian at the University of Miami, has made. There is an accessibility course, there is a data feminism course, and all of them are free and open source. So you just go there. You don't even need to register for the platform. You can just watch them. Now. What you will not get there is what I would say the most important part of the learning process, which is the experience, the practice. You need to put all these ideas in practice. These are essentially just lectures and tutorials. They're also software tutorials. We have intros to R data wrapper and many other tools. But that will be on you, that will be on the viewer after you watch all these, take notes, study the courses, then you need to get started right away producing content. But at least I wanted to lower the barrier of entry by telling people, which is another thing that I usually tell my students in classes. This is not magic. Everybody can learn and anybody can learn. And if you Learn you're going to become a better thinker, but also a better communicator. [00:39:11] Alex: I love it. Yeah, I've really enjoyed looking through a lot of the materials, and it's definitely informed a lot of the way that I teach about visualization to my students. [00:39:19] Alberto: Appreciate that. [00:39:20] Alex: So I'm kind of curious though. You've been doing this now for 15 years and you talk about the importance of the audience and their understanding. A lot's changed over 15 years. So I'm curious, how has that been reflected in your teaching? And how do you find that your teaching has had to evolve to keep up with the now growing importance of data and AI in today's society? [00:39:40] Alberto: Yeah, I mean, I try to keep my university courses fresh. I mean, there's a core. Things that don't change, like the ethical part of things, the reasoning part of things, the importance of spending time and paying attention to what you're doing. Sort of like the importance of being deliberate about the decisions that you make. All that has not changed. I would say that they have become the. That part of my classes has become more important throughout the years, precisely because of the rise of easy to use tools, not only AI, but before that, point and click tools and programming languages that sort of make the process easier. And thinking about R and G plot and the entire tidyverse, when I discover that, by the way, it's like, God bless Hadley Wickham. I mean, it was amazing. It was sort of like a discovery from how easy it was to create visualization through coding. I don't like to code, but that was so easy. It came so naturally to me. At the same time, lack of friction comes with some downsides, which is that you may want to move too quickly to the outcome. And this is something else that I teach my students. It's not about the outcome, it's about the process, because the process will guarantee the outcome. If you don't engage in the process, you cannot ensure that the outcome is going to be good. So again, regardless of what tool you use, a programming language point and graphical user interface tool, or a generative AI, the process doesn't change. It doesn't change. It's the same. I have been updating the tools that I use in my classes. I emphasize some part of the teaching more than others. I try to incorporate modern examples every semester. But again, the core has not changed that much. [00:41:22] Alex: So what do you think comes next? What's this in store for the future of data visualization? [00:41:27] Alberto: I have no idea. I mean, I think that generative AI is going to change the landscape quite a lot. I think that in the one hand it has a lot of potential to be beneficial, but not in its current form. I think that the underlying technologies behind generative AI, machine learning, deep learning, neural networks and so on and so forth, extremely powerful tools, extremely applicable to visualization and many other areas. Areas, but in their current instantiations, meaning the most powerful technologies belonging to companies that don't have the public good in mind that can be extremely detrimental to society because the incentives of those companies is not our well being, it's making more money and eventually they will need to make more money. And so I think that we are in a world in which all these companies, I'm thinking about anthropic, I'm thinking about ChatGPT and OpenAI, etc. They are implicitly, explicitly, inadvertently or deliberately engaged in a process of cognitive capture. They want to capture as many minds as possible to make people dependent not on the technologies, but on their instantiations of these technologies. So people will forget how to make things in a different way and eventually raise prices. That's my fear in the future that we are going to see sort of like a much larger axis inequality and cognitive inequality. And I want to fight against that in some. Again, the problem is not the technology, it's the structural problems, the structural instantiations, current instantiation technology. But I dream about a future in which we will have smaller pet specialized models that work only in our computers to do very specialized things. I dream about a future in which people, even people will be able to sort of overcome sort of like the structural problems of all these tools by using their judgment. So I dream about a future in which people will be more judgmental about the tools, but also judgmental about their ways of using tools. But I honestly cannot predict where all this is going. [00:43:44] Alex: One analogy I kind of give is that when the cell phone came around, it really lowered the barrier to usage for or making videos. Suddenly everybody could become a filmmaker. And I feel like a lot of the AI tools that are available right now are likewise lowering the barrier to becoming a data scientist and a data visualizer. Everyone can visualize their own data at the moment. Do you think that that's going to increase the literacy of everyone can make it or is it going to happen? [00:44:12] Alberto: It may happen, but it will not happen on its own. Right. It's like because the illusion that all these tools create in your mind is that you are creating the graphic and you're not, you're not there's a cartoon that I just published in my newsletter in the Open Visualization Academy newsletter. On the first panel, there is a man sitting on a table at a restaurant. And there is the waiter. The waiter comes and the man sitting at the table says, please bring me a pizza. And then the waiter brings the pizza. And then in the last panel, the man reacts. Look what I made. And the cartoon is the AI approach to design. I fear that that will happen. I fear that delusion that you are making something with generative AI without guiding it, because it is different. For example, if you use again, gen AI, as again, a natural language interface, the same way that you will use a graphical. When you use a graphical user interface tool, you guide the tool. You click on something, you change the axis, you change the colors, you do these, you do that. It's a very iterative process. Now, if you use the gen AI the same way, right? If you prompt the tool, now put this variable on the x axis. Now this variable on the Y axis. I don't like it. Change this, change that, change this color. If you use it that way, that's the right way to use it. But again, using these tools this way requires prior knowledge that the tool itself on its own will not give you. You need to develop it, maybe with the tool, with the help of the tool, but also with the help of reading books. I mean, we need to read books. We need to watch courses in the Open Visualization Academy or whatever, right? There are many other places where you can learn about all this stuff, all that matters to avoid that delusion that again, the big tech companies are fully involved in sort of like spreading. Spreading this delusion that this is all magic, that now you can create music right, no you can't. It's the tool that is making it for you. It's not you who's creating it. So you need to take ownership of the tool. This is true, by the way, both with gen AI or with previous tools. I mean, there are tons of charts that are really bad, made in Excel or ggplot or Tableau or whatever. If you don't make the charts deliberately, if you get carried away by the tool, then it's not you who's creating the graphic, it's the tool. Something that I tell my, that I joke with my students every semester is that as a reader, I deeply dislike being able to tell that a chart has been made by a tool or being able to tell what tool was used for a specific chart. Because I want to get a sense that there is a human being behind that, making choices to make my life easier. That's part of the social contract of visualization design. It's like I'm going to create this graphic to illuminate you, to bring you more understanding, and you need to sort of like convey that through the design decisions that you deliberately make when you create a chart. [00:47:18] Alex: I love that. So today's era and crises is not a new formulation of it? No, it's sort of like [00:47:24] Alberto: it's just as a scale. It's like an expansion of challenges that we already had. The difference, though, again, again, is sort of like the structure is the system is the big tech again. I dream about a future in which we will take ownership of all those tools in which all these companies crash in the stock market. I think that many of them should disappear because they are detrimental to society, but the technology will remain because the technology is useful. Machine learning, deep learning, neural networks, whatever, all these things are extremely useful if you know what you are doing. But if they are publicly owned, if they are open source, if they are specifically targeted to specific purposes rather than general purposes, and so on and so forth. [00:48:12] Alex: It's such a pleasure having you here. Thank you. Thank you so much. We look forward to your talk tonight at the Data Is Our Finalists competition. [00:48:21] Margaux: Thanks for listening to UVA Data Points. More information can be found @datascience virginia.edu. and if you're enjoying UVA Data Points, be sure to give us a rating and review wherever you listen to podcasts. We'll be back soon with another conversation about the world of data science.

Other Episodes