← Harvard Data Science Review Podcast
Harvard Data Science Review Podcast
In God We Trust: Everyone Else Must Bring Data or Liberty
00:32:01
Loading summary
Transcript52 lines
- [00:01]
Ariane Winfrank
Hello and welcome to the Harvard Data Science Review Podcast. I'm Ariane Winfrank, a producer for this podcast, a data science undergrad student at Washington University in St. Louis, a former student of Liberty Vittert, and most importantly, the guest host of today's episode. I will be joined here today by co host and editor in chief of the Harvard Data Science Review, Shelley Mae. After the positive reception of our listeners Question special featuring Shao Li Meng in August, we decided to commence the new year with an exclusive interview featuring one of our most esteemed guests yet, HGSR's own Liberty Vitter. Today, we will get to the bottom of how a woman with so many hats found herself in the exciting world of data science. Get ready to uncover captivating stories, glean valuable undergrad advice, hear Liberty's perspective on generative AI and much more on this month's installment of the Harvard Data Science Review podcast.
- [00:59]
Shelley Mae
First of all, happy New Year. And I'm so happy to have a chance to grill my co host, Liberty, because she did that to me last year. I've been waiting for this opportunity for a whole year. So, Liberty, here is the first question. You wear so many hats. You're a statistician, professor, journalist, cooking show host, podcast host, doc series host, books and journal editors, and so much more, such as working with the UN refugee agency. How did you manage to do so much? And what are a few common principles you hold yourself to as the most versatile data scientist I have ever met, other than always pour yourself a glass, since we also know you are a wine connoisseur.
- [01:43]
Liberty Vitter
Well, I learned from the best, Shelley, to always pour yourself a glass. That came from you. So I appreciate the best advice I've ever gotten. Thank you. And it's fun. Funny you asked this. It is something that you said when we interviewed you for the podcast, and I never could have said it as articulately as you did, but it's that one thing leads to another. Especially at the beginning of my career, I said yes to absolutely everything, even things that I thought, what on earth could this do? Or why would I do this? And some of those things, not all, some of them lead to nothing. But a lot of them have led to some of the coolest things I've ever gotten to do. And you never know which thing is going to lead you where. And you know, I mean, Shelley, that's how we met. I was sitting at a conference that I frankly, really didn't want to go to. It was like 8am on a Sunday or something. And I'M sitting there and I'm sitting next to you, and we'd never met before, and Robin actually came up to me and said, oh, I heard you have a cooking show. And you looked at me and you went, cooking show. And I mean, that's how this podcast. This podcast never would have happened unless Robin had seen me on a cooking show. So you just. You never know what's going to lead to the next thing. So I would especially encourage any young data scientists or anybody young and in anything. Ari, you included that. To really have that energy, surround yourself with cool people who are doing interesting things, be interested in what they're doing, which, by the way, I've seen incredibly from Shelly. He's always interested in what everybody else is doing, and you'll end up doing really cool things and getting really cool opportunities.
- [03:17]
Xiao Li Meng
I mean, what you said is exactly right. The only reason I even worked on this podcast with you was I just emailed you and was like, hey, are you doing anything cool? And you're like, actually, I am doing something cool. Do you want to get your hands on it? And now this is my favorite project that I do. And so leading into that, as academics, we're always learning and growing, and obviously you do so much here. But what has this podcast taught you?
- [03:42]
Liberty Vitter
This podcast, I agree with you, Ari. This is one of my favorite things that I get to do, because what's so cool about data science and what I really encourage young people who are interested in data science about is that data science touches all of our lives, no matter what you do, Whether it's. And we've learned this from the podcast, whether it's the art world or the criminal world or. I mean, we've done guns, we've done marijuana. I mean, we've done everything you can possibly think of that data science touches, and it allows you to have this really vibrant, dynamic career where as long as you know the basics of data science, you can get involved in so many different fields and you can be interested in so many different things. And I think the Harvard Data Science Review podcast and journal are perfect examples of that. If you look through our podcast episodes, it's on everything you can imagine. If you look at the journal, it touches every different aspect of life. And so one of the joys of the podcast is, what do I want to learn about this month? Or what do our listeners want to learn about this month? And it's really one of my favorite things I get to do.
- [04:51]
Shelley Mae
Well, thank you, Liberty. But I do want to clarify. We did not do Marijuana, we talk about marijuana just to be sure that the audience get it right.
- [05:01]
Liberty Vitter
We know there's always a first, Shallie. There's always a first. There's always time.
- [05:07]
Shelley Mae
Okay, all right. Okay. Before you got us into more trouble, let me ask a question. I know you do extremely well. You are known for making understanding statistics accessible to a wide range of people. There are many people who think statistical concept, data concepts are very difficult. How are you able to explain such difficult concept to such a wide range of audience?
- [05:29]
Liberty Vitter
I think there's two keys to it, and one I think is not going to be very popular to say on this podcast, but I'm going to say it anyway. The first is analogies are king. So if you can come up, you know, if you're talking about, you know, I don't know, regression to the mean, if you can come up with an interesting example that relates to whoever you're talking about, whether it has to do with politics or drugs, healthcare drugs, not marijuana drugs, whatever it is that you can come up with, if you can come up with analogies that make sense in people's fields, they can really understand things better. The second is that I am willing, and I know this is going to be controversial, but I am willing to sacrifice technically correct, perfect correctness for understanding so that sometimes you have to skip over the details, sometimes you have to skip over the technical correctness of something for people to understand. And I would rather have them somewhat understand the statistics and somewhat appreciate the data and the truth and potentially sacrifice some of the correctness rather than have them just go, I don't understand this at all.
- [06:36]
Shelley Mae
Well, I can say why you are saying this could be controversial because I would know. Some of my colleagues might be saying, you know, but I know exactly what you mean. I guess usually what we talk about here is you find a good knowledge, you find a way to explain the things which may not be precise. It's not wrong, but it's not precise. It does not give all the nuance, the angles, but it get the essence cross. I think that's, that's the essence of.
- [07:02]
Xiao Li Meng
The bad, best communication leading into that, actually how I always explain to people my job, because I think even explaining to people the essence of being a data scientist is really hard. And so I always tell people, like when you see the commercials, let's say 9 out of 10 dentists recommend this toothpaste, I'm the one that got the nine out of 10, and people think that that's really great. So on imperfect analogies, people say, that learning data fluency is a lot like learning a whole other language. But we took a look at your resume, and I did not know that you are competently fluent in five different languages. Is there any intersection between learning and speaking in different languages and, like, kind of learning and speaking data science? Do those two subjects collide at all?
- [07:56]
Liberty Vitter
Oh, that's so interesting. First, I'll say I'm really happy, Ari, that you do remember something from my class three years ago, because the Colgate crest, nine out of ten dentist examples, was mine. So I'm glad to know that at least one of my students remembers one thing from my class. But that's an interesting thing. I actually found learning another language really hard. And I also found learning statistics really hard. I loved it, but it was hard. Everyone says, like, oh, did math just come naturally to you? Actually, in eighth grade, my eighth grade teacher in high school called my parents in for a big meeting and said, liberty's just not smart enough to graduate high school. She's not good enough at math, and she's just not smart enough. And it was a private academic school, and they said, she's just not smart enough to be here. So to all these parents who might be listening or kids who may have found things hard, it doesn't necessarily come naturally. If you think sort of about basic math and reading in school. If a kid can't read, people don't say, oh, well, that's fine. We'll just let them not learn to read. They figure out how to teach them to read. Whereas if a kid says, I'm not good at math or statistics, parents will go, oh, well, my kid's just not good at it. I wasn't good at it either. And we don't take the time to really figure out how people should learn because people learn differently. And so I think my big thing is that if you're a parent listening or a kid who struggles with this, that I think everyone can do this. You just need to figure out how to learn it and that people should really do that for their kids because it's an important thing and it does not necessarily come naturally, as neither did languages. I mean, Xiao Li, did Chinese and English. Did English come really easily to you?
- [09:38]
Shelley Mae
I wish. I'm still struggling, as you know. Well, I share my writing with you. I'm still struggling with all these things, like, always missing, you know, where should be the. Or should it be the A? You know, all these articles, where to put the what, Which I just had no idea. But when you're not Native speakers. It's just hard. Right. And so.
- [10:01]
Liberty Vitter
But you're funny. I don't know. That's what everyone says is the mark of someone who truly understands their second language, is that they're able to be funny in that other language and you're hysterical. So I. Clearly, you understand your mastery of English has leveled beyond anything that anyone could ask for in understanding.
- [10:19]
Shelley Mae
Well, I guess probably people just saying that my English reads funny because it doesn't really have the right tone, but that's fine. I take that as compliment. But I want to jump in that you were saying that you had trouble to study mathematical. Now I need to let the audience know, Right. This is from someone who was a math major from mit, so let's just get the record straight. But I also know that you concentrate in political science when in doing mathematics. I think that's a wonderful combination. Probably explain part of the. Your incredibly diverse interest. Right. Because most people think about political science as mathematics. They don't necessarily, you know, talk to each other, but obviously both are incredibly useful skills. Speaking of the language, I think this is actually a perfect time to brought up this most frequent question from Instagram. Okay. Because we did ask Instagram, right. About asking, you know, answer questions. Because this is obviously related to the whole language issue. This whole generative AI is all coming from large language models. So you know how the two things have become so intertwined with each other now. So the question is, as a professor, how do you feel about ChatGPT and other generative AIs in the classroom?
- [11:37]
Liberty Vitter
You know, I'm just in the process of revising my syllabus. I feel like I made a mistake. The general decision at the universities has been sort of the general overall one has been to say, no, ChatGPT, you can't use it. And I feel like I really made a mistake because last year I went along, I picked the path of least resistance and said, fine, I'll make it the equivalent of plagiarism in my syllabus. And I thought, I was thinking about it this fall and I thought, what an idiot I was to do that, because people are using this in their jobs, people are using this in work. And what a disservice to our students to not have them be using this in what they do. And if that takes extra work on the professor's part of figuring out assignments that you can't just put into ChatGPT and get the answer, then that's on us. And we need to take the extra time and effort to come up with assignments that really do stretch your mind and utilize large language models and stuff like ChatGPT and helping our students utilize something they'll be using in the real world and to further and enhance their work rather than say, no, you can't use this incredible tool that everybody's already using in the workplace. So I am happy to say I was dumb and I regret my decision and I am changing it for this semester. So I don't know what that means for my students. But I'm coming up with new assignments as we speak.
- [13:05]
Shelley Mae
Let me actually jump in to share with you a story that someone just sent me a newspaper clip most recently to show that, you know, that kind of reaction at the beginning is probably all natural and you're not dumb at all again. And somebody sent me a news clip. This is from years ago. There was a group of mathematical professors was doing protest on campus, you know, with signs that basically is against the calculator and because they were worried about calculators going to, you know, pollute all the young minds, you know, how, you know, they're kind of now we. Looks back looks silly, but that's how we react when the new technology comes.
- [13:47]
Xiao Li Meng
As a student, I'd like to thank you for your amendment of your syllabus because I feel like we have all of these tools at our fingertips. And so sometimes classes that treat generative AI like plagiarism are actually hurting us because they end up being kind of base level, they end up being kind of shallow. And so I found that most of my classes that let us use generative AI as like a tool to help us are much more in depth. And like, I feel like those classes prepare me more for the workforce. So I think from a student perspective, two thumbs up for sure.
- [14:25]
Liberty Vitter
Thank you.
- [14:25]
Xiao Li Meng
That said, I'm calling all of your students from last semester and apologizing. So next, as someone who has achieved so much in such little time, you know, we've talked about generative AI. Data science is going skyrocketing upwards. Where do you see yourself in the next 10 years? And where do you see the enterprise of data science as a whole headed in the next 10 years?
- [14:53]
Liberty Vitter
You know, it's funny you say achieve so much. I don't know if this happens to everybody, but I feel like I've done nothing. I was just having breakfast with my dad at Chick Fil A this morning. I had a peppermint milkshake, which they only are there through Christmas, so I need to get in as many as possible before the holiday season. Is over. And I was talking to him about how I feel like I need new challenges and my career's stagnating and I need to do more things. So, you know, I think the one thing is that for anybody that's interested in trying to do stuff and excited about things, that everybody always feels, you know, I don't know, Shelley, how you feel, what's your. What's in your next step and next part of your life? Like, that's one thing I always watch with Shelly. He's always doing new things and coming up with new things, and that's how I want to be, is that you're never sort of just sitting there, content with what you're doing, and you want to, you know, get excited about new things. So I don't know, what's the next step? I think I'm going to start saying yes to everything again. I think I almost stopped for a little while because I was getting so busy, and now I think I made a mistake. And I think my first step is going to be to start saying yes to everything. Someone actually just asked me to give a talk, and I thought, oh, God, I don't want to travel there and do that. Nope, I'm doing it. So I'm going to start saying yes to new things. And I don't know what the future holds, but I'm excited to find out because I'm ready for my next. I don't know, my next life as a data scientist with always having that background and that root in data science and as a teacher, which is my favorite thing to do.
- [16:22]
Shelley Mae
Well, I will say that many of people who is listening to this podcast will be very happy to say you're going to say yes all the time, but be careful what you just said. I'm quite sure your invitation list is only much longer than mine. Even I have, you know, trouble to keep up with. Of course, I'm a lot older. That's a problem. And, you know, just be careful. Just be careful when you say too many yeses. But speaking of your achievement, I do think there's one thing you have achieved that many of us have felt which you may or may not realize. So I know this is a little bit of a serious question, so I want to ask you, how do you achieve that? You know, you have written four of being featured by a wide range of media outlets known for their diverse political inclinations, including BBC, cnn, Fox News, News Nation, pbs. You know, people will say these are. Have very different, you know, ideological perspectives. Right. Just to name A few. I know you have done a lot more. If you look at the cv, it's really a long list. In an increasingly divisive society, unfortunately, how do you manage to be appreciated and respect by people with very different ideological perspectives?
- [17:41]
Liberty Vitter
You know, I know I keep saying this. I think this podcast is making me realize how much advice I've gotten from you, Shelley, in doing all this stuff. One, I feel like I really always try to have a backing in the data. You know, whenever I'm writing an op ed or an opinion, I really try to feel I could defend it either direction. I mean, you always know that you can make the. What's the thing? If you torture data long enough, it'll tell you anything you want it to say.
- [18:10]
Shelley Mae
If you torture data long enough, it will confess.
- [18:12]
Liberty Vitter
It'll confess. There we go. So I feel like I tried to make sure that I can make the argument either way with the data. And I include that in any opinion piece I write. I may take an opinion at the end, but I include the other side in it every time. And I believe me, I've almost made some big missteps and my grandfather always used to say never react. And I've almost made some really big missteps by reacting to an issue or something that I'm going to write about without really thinking it through and without talking to people that are a lot smarter than I am about whether I should write it or not or say it or not. And so I think I've gotten lucky enough to have people around that I do ask before I say stuff and get their opinion on it first. So always have people around you that are smarter than you are that you can really talk to and have them understand. Because I've almost made some big mistakes and thankfully I haven't, knock on wood, I haven't yet because I've taken a step back and let my emotions come out of it and asked people that are smarter than I am about what they think about it.
- [19:19]
Shelley Mae
I think that's just such an important piece of advice. I know it's many of us trying to do, but it's not easy, right? What you're saying is really you got into a positive cycle because you do well, you get more people with different perspectives around you. So whenever you are getting to issues, you have people with right. Different perspective consult, in which case we'll first help you to present, you know, data science, much more balanced view. And you got into the, you know, the right part of cycle. Unfortunately for a lot of us, and certainly these Days. You know, I, I really dislike the notion called a personalized news because what happened is you just keep hearing what you want to hear and then you end up in a ways feel like, you know, how could anything else happen? Because this is the world I live in. Right? So I think, I think your best advice here is really surround yourself with a lot of people, but a lot of people with different, different perspectives and then going from there. Well, thank you for that wonderful advice to all of us.
- [20:19]
Xiao Li Meng
I actually remember my favorite class period with you was probably when you go through, I think it's maybe the baby boxes example where you talk about, you wrote this, like, really insightful article. And then towards the end of the class period and you're like, and then I calculated all of this and then we went through all of this and then someone wrote an article refuting me. And that was devastating. And I still tell people about that all the time. Because if a professor can, like, stand up in front of 200 kids and be like, and this is what they said about me.
- [20:54]
Liberty Vitter
I think the headline was, is Liberty Vittered actually this stupid? I think that was the headline.
- [21:00]
Shelley Mae
Wow.
- [21:01]
Xiao Li Meng
It's just like, it's so important to be able to look at it and be like, hey, so this is why this article that was written about me was maybe right. Or this is where it wasn't right. And I think it's really excellent that you do that. I think that's super powerful, not only for your students, but just people in general. Like you can tell by talking to you that you're okay accepting when you're not right.
- [21:23]
Liberty Vitter
I've been wrong so many times, I can't even count. So get in line if you want to come up with times I've been wrong because there's been many.
- [21:33]
Xiao Li Meng
Well, on times that you have been right, we are talking about. This is our fourth year of the Harvard Data Science Review. No one has written an article about us saying that we're stupid. So that feels like a win, as we are in the new year. What has been your favorite or one of your favorite Harvard Data Science Review episodes so far?
- [21:57]
Liberty Vitter
So I've had two. The first, and still my favorite and also our fan favorite was the interview with Shelley. Because, and that was Ari's idea, was because to have him not have the questions beforehand and really just answer off the cuff was fascinating to see and to learn the way he was thinking about things. So I would highly. It's called what is Data Science? And I would highly. I would highly encourage our listeners to Go listen to that if they haven't, although I'm sure you have, because it's definitely our favorite. The second. And this, I loved this one because I really didn't think this is what was going to happen. It was our to drink or not to drink episode. And first of all, thank God the advice was to drink because I don't know what I would have done. But I was really surprised because I feel like we hear. I kind of have a good guess of what people are going to say in the podcast or what angle they're going to take. And I was shocked by both of these people. Sort of even one that really had just written an article, a real big article saying that alcohol was bad for you, really ended up saying, well, if you have a couple drinks, if you drink responsibly, it's probably okay for you. And so it was the first time I sort of heard a really nuanced discussion of something like that. And I was surprised by what the happily surprised by what the answer was. And it was just such a fun podcast episode because it was two people with such different views who really came together at the end to agree on something which you see so rarely.
- [23:37]
Shelley Mae
Well, thank you, Liberty. And but I had to clarify otherwise sounds like this whole episode that we planned, in the end you were keeping talking about me because it's really about you. But thank you for all these very nice words. And but I do have to say I deny that we never done marijuana, but we have done wine. That is true. So that one like that one, I can clarify. And I'm just going to, you know, this, this conversation obviously can go on forever, but we will. We're getting to the magical one time and you are the one always come up with these wonderful magic wands. So you have to wait for whatever magical wand. You know, we, we will ask you. But before I do that, I do want to ask you because you are the one. I also take this opportunity to really thank you for the last three years for your incredible dedication to this podcast. Together with Ari, Tina, Rebecca, the whole team had come up with all kinds of wonderful, wonderful topics. I can tell the. I can tell the audience. My job was much easier because I basically just come in and talk and all the topics coming from. Coming from you guys. Of course, if anything wrong, you guys get blamed as well. So I'm just trying to be clear here. Okay. But seriously, what are the one absolutely you would love to do this year coming to 2024 and any particular guest that you like to have Ooh, this.
- [24:58]
Liberty Vitter
Is making me want to go look at our wish list and remind myself of what episodes we have coming up. You know, I think I would. I think I definitely want to do a follow up on Large language models in ChatGPT. You know, we did an episode sort of right at the beginning of all of this when people were really trying to figure out what's going on. And I think there's still a lot of, you know, you read these doomsday articles about how AI is going to destroy everybody. And I think it would be really interesting, now that things have settled down a little bit around it, to have a discussion and whether we should be scared or whether things are okay, whether we, you know, maybe we should be terrified, but maybe we shouldn't be. And I'm thinking, you know, I thought we had a spectacular conversation with Steven Pinker over it, and I think it would be really cool to have him back on to sort of, now that the dust has settled around this chatgpt and generative AI to see what he thinks now, a year later, I think could be really fascinating when where we're going to be, especially with all the Hubba Balloo with OpenAI recently with Sam Altman and what's going on with how people are thinking about the future of AI. So I think that'd be off the top of my head. That is my first choice of episodes to have some time this spring.
- [26:19]
Xiao Li Meng
So, as this podcast only undergraduate student that's not big and famous and incredible yet, what advice do you have? Liberty. For us, people that are just starting out and want to do big, great.
- [26:31]
Liberty Vitter
Things, my biggest advice would be to take risks with what you can try to do. You know, Ari, you even talked about it. You emailed me and said, like, you know, what can I do? That's interesting. And frankly, you caught me at a moment where I get emails like that all the time. And you, first of all, I loved you as a student, but you caught me at a moment where I had an opening, I needed somebody, and it was perfect. And unfortunately, there's times where students email me and ask that who are great students. And I just don't have anything and I forget and that's it. But there are students who keep emailing, and that's what I think students need to do, whether it's to professors or to work or to whatever it is. Cold email people. Cold call people. 99% of them are not going to respond to you. The other half a percent are going to say they don't have anything. For you. And that 1/2 of 1% might have an opportunity for you. And that's how so many of my opportunities came was just being relentless about emailing and calling people and being rejected and not caring. And that half of 1% is potentially really cool things, by the way, that half of 1% can sometimes lead to nothing, but a lot of times they're super interesting things. So it's just to really put yourself out there and you'll get rejected a bunch. And it's fine.
- [27:50]
Xiao Li Meng
You heard it here first, folks. Hire Arianwen Frank.
- [27:56]
Shelley Mae
So I guess I can summarize what you're saying is keep asking, but whenever people ask, you say yes.
- [28:02]
Liberty Vitter
Exactly. And by the way, it's a great opportunity to me when I said yes. Ari's been incredible and done so much for the podcast. So it's also been a wonderful thing that I said yes too, on the other side for me and for us.
- [28:15]
Shelley Mae
Thank you very much. Now come to the magical wand question and this one Ari prepared. If you could accurately quantify anything that is currently unquantifiable, what would it be and why?
- [28:29]
Liberty Vitter
What do I want to quantify that's unquantifiable? You know, I think one of the biggest discussions that's going on right now, or sort of in the little world that I'm in right now, is on the political side of both sides are basically calling doomsday if the other side wins. The Democrats are saying it's the end of the world if the Republicans win. And the Republicans are saying the biggest threat to our democracy is the Democrats. Everyone's calling for doomsday and everyone talks about how trust has fallen and no one trusts anything anymore. I'd like to really understand public trust over time. I'd like to know, has it really changed? Are we really the most divided we've ever been, or is this just another ruffle in time? So I think to really understand what's going on and how at risk we are for democracy falling, I'd really like to understand the trends in public trust over time. And are we at our most divided? Should we be really scared about the fall of democracy or not?
- [29:39]
Shelley Mae
That is truly a fascinating answer. Seriously, that's a really great answer. Yeah. How do you measure trust even during the kind of a normal time now, this divide time. Right. And who's doing that? And how do you trust people's evaluation of these trust? Because everybody talk about doomsday. I think that's that itself deserves a great PhD thesis in political science with the data Science spin.
- [30:02]
Liberty Vitter
Seriously, Ari, it's all on you, Ari.
- [30:06]
Shelley Mae
That's your topic. Okay, but. Well, thank you. Your measure of trust also reminds me a quote. Actually my apology. I don't remember who said this is in God we trust everybody else. Bring data. That's the. Which is pretty good one.
- [30:22]
Liberty Vitter
It's a pretty good one.
- [30:24]
Shelley Mae
And bring Liberty. I think that'll be even better. Okay, but seriously, that we need to wrap up this episode. And thank you again, Liberty, for being such a great co host. And honestly speaking, the Harvard Data Science Review podcast will not exist, or certainly not in this form without your tremendous dedication in the last three years. I'm certainly looking for many, many more years. And I know now you cannot say no to me because you said you're going to say yes to everyone. So I hope at least I'm one of everyone's. Like, I don't even have to claim anything special. Just say say yes to me.
- [31:09]
Liberty Vitter
I will say yes.
- [31:10]
Shelley Mae
Okay. And surely I will open my wine cellar anytime you come to Boston.
- [31:16]
Liberty Vitter
As always, champagne, it is time.
- [31:19]
Shelley Mae
Yeah. And thanks again.
- [31:23]
Ariane Winfrank
Thank you for listening to this week's episode of the Harvard Data Science Review podcast. To stay updated with all things HDSR, you can visit our website at HDSR, MITPress, MIT.edu, or follow us on Twitter and Instagram HDSR. A special thanks, our executive producer, Rebecca McLeod and producer Tina Toby Mack. If you liked this episode, please leave us a review on Spotify, Apple or wherever you get your podcasts. This has been the Harvard Data Science Review. Everything, Data Science and Data Science for everyone.