Loading summary
Transcript136 lines
- [00:00]
A
Foreign. Welcome back to the AI Policy Podcast. This week's episode is going to be a little different. I'm Matt, a research assistant here with the Wadhwani AI center, and today I'm talking to Greg about how to have a successful career in AI Policy. We decided to do this episode after hearing from some members of our audience interested in career advice.
- [00:30]
B
And.
- [00:30]
A
And as someone who has benefited quite a lot from Greg's career advice myself, I am super excited to share all his ideas with our audience. So, Greg, thank you so much for taking the time to do this, and welcome back.
- [00:42]
B
Yeah, I'm really excited to do this. I mean, like, I have a pretty fun and interesting job and, like, hosting this podcast is one of the fun and interesting things I get to do. It's kind of always amazing to me that people actually care what I think on various issues or are willing to take the time to read what I write. And it's also been amazing to, like, hear from listeners literally all over the world for this podcast. I mean, like, the audience for this podcast includes not just, like, senior officials in the US Government, but senior government officials from around the world, academics, industry folks, et cetera. It's all amazing. And one community that listens to this podcast is young career professionals who are, you know, either early in their career in AI policy or thinking about a career in AI policy. And I get a ton of emails from these folks. And so rather than have this conversation, you know, 500 times, I thought, let's just have it once and record it so I can at least get the stuff that I say every time out. And so this, as I, as Matt said, this is going to be kind of a different podcast, and it's gonna involve, you know, some stories from my life that I thought included helpful career lessons that I've tried. But, Matt, you know, you can. You'll know when you. When you hear my answers to some of these questions. I'm sure that this is similar to the, you know, the time I took you out to lunch when you started working at CSIS and some of the stuff that I try and advise folks. So I hope that this is useful. You know, if you are a senior government official, probably gonna be less useful to you, but maybe you'll find it interesting anyway.
- [02:27]
A
Yeah, And I was one of those early career professionals sending you emails, Greg. Remember that?
- [02:32]
B
Yes. And if memory serves, I foisted you off on a different research assistant who works here, Isaac Goldston. But the story has a happy ending now. You work for us.
- [02:42]
A
That's Right, well, let's get into your background then.
- [02:46]
B
I think we should say sort of like the flow of the conversation is going to kind of go in three stages. So the first is, you know, a little bit about my background and sort of how I got to where I am. And then the second is sort of what I would call generic advice for succeeding in some cases, like any kind of job, in some cases a little bit more geared towards, like, think tank and policy world. And then it'll conclude with what I think are, like, the specific stuff about AI. But my, my sense is that most of the advice for how to be good at AI policy would also apply to how to be good at many other kinds of jobs.
- [03:28]
A
Yeah, that sounds right. So, yeah, let's jump into that first bucket of questions, which is your background. Can you just tell us how did you get to where you are today?
- [03:37]
B
Yeah. So I grew up in Kansas. I lived across the street from a soybean farm and about a couple blocks away from a cow farm. I was not a farmer, but people I went to school with were farmers. And it was kind of interesting because, like, I lived in the outskirts of the outskirts of Kansas City, so, like, where the suburbs meet the farm country. And Kansas City has grown a lot since I was there, so now the suburbs have eaten everything. But, you know, I grew up and my family has, like, a history in the technology industry. My dad did stuff in the computer industry. My brother is a, like an astrophysicist and computer programmer. And so I was always interested in science and technology and definitely thought that that's what I was going to do when I grew up. I definitely thought I was going to be an engineer for most of my life. But then in high school, I really fell in love with the debate team, and I was a huge debate nerd.
- [04:52]
A
I can see that for you.
- [04:53]
B
Yeah, no, it, it, it checks, right?
- [04:55]
A
Yeah.
- [04:57]
B
And debate in Kansas is kind of this just shocking cultural phenomenon. So of all the kids in America who do high school debate, I mean, I don't know what the population of, like, high schoolers in America is, but the population of America is north of 300 million, and Kansas only has a population of 3 million. So 1%, the population of the whole country. But in terms of high school debate, one out of every three high school debaters in the country, at least when I was in high school, is a Kansan. There's just a really strong debate culture in Kansas. And debate was a huge deal at my high school. And then the way Kansas does it is like, half the year is debate, and then the other half is competitive speech and drama. So I got like, crash course in both the, you know, research side of public of debate, because what I did was called policy debate, where you actually had to, you know, go read about whatever topic you were going to be arguing about. You would, like, read reports or read news articles, read speeches, and then you would. We would call it cutting cards, which is to say we would literally, like, print out the article, get out scissors, cut out, like, the paragraphs that were useful to us, tape them to a new sheet of paper so that in the debate round we could say, hey, you know, I say that this move is going to be bad for US Foreign policy. But don't take my word for it. You know, some measly high schooler, here's like a fancy report that claims, you know, what I'm claiming. And so I knew what CSIS was when I was in high school. Like, I was reading CSIS reports and other think tank reports when I was high school, and I really just sort of caught the government and public policy bug and was really appealed to that. And of course, what was especially appealing to me was like, the, the science and technology policy part of it all.
- [07:03]
A
And so I think, I think that's pretty unusual. I think most people don't know what a think tank is ever, maybe. And so you were in high school learning about what think tanks are and reading the research.
- [07:16]
B
Yes, because think tanks produce a lot of policy research. And when you're doing what I was doing, which is called policy debate, you know, that kind of stuff is really helpful. And, you know, there's community. Like, there's actually a lot of other former debaters who are in AI policy, like Jeff Ding, who's a professor at gw. He was a former debater, very successful. Mike Horowitz at UPenn, who was previously a deputy Assistant Secretary of Defense doing AI policy, among other things. He was a former, very successful, successful debater. So, like, you know, we debate nerds, we're represented in the AI policy community. But when I, when I think about, like, what I really got out of that debate experience, other, other than just like, learning that think tanks were a thing that I might want to do with my life or that policy was a thing that I might want to do with my life, like, a few things kind of stand out to me. The, the first is, you know, how to do research, of course, but in debate and in policy debate, you're constantly switching sides. So you will argue one side of A topic, for example, that when I was a junior in high school, the topic was United nations peacekeeping operations. So I would argue, you know, united nation peacekeeping operations are great, and we should definitely do more to support them. And then literally, you know, 90 minutes later, I'd argue the other side of the issue. And so the arguments that I'd just been building up, now I have to tear them down. And that is like, a phenomenal intellectual exercise, you know, to competitively go against both sides of an issue. And, man, I mean, like, this was a huge part of my life, I think. I went to, you know, a Debate round is 90 minutes. A standard tournament is five rounds. And then if you make, you know, octa, finals, quarterfinals, semifinals, finals, et cetera, it goes on even more than that. And I was doing, like, 10 debate tournaments a year and 10 competitive speech tournaments a year, every year, four years of high school. And so that was just a lot of practice in public speaking, a lot of practice in arguing stuff, both in the research side, like, upfront, you know, getting an arsenal of evidence ready for the round, and then also in the round and having to come up with stuff on the fly. And that was. That was really, really helpful for me, that exercise of, like, you know, creating arguments in frontline. So, like, people often ask me for public speaking advice because, you know, I'm. I'm a loudmouth public speaker, and people can hear me, you know, from. From the back of the room. I guess I can.
- [10:00]
A
That.
- [10:01]
B
Yeah. And the, like, my number one thing is like, well, are you practicing public speaking? And they're like, no. And I'm like, well, I'm shocked that you're not getting better. Right. Like, show me in your calendar when you're supposed to practice this sort of thing. And it was just kind of this. This wonderful experience of, like, starting out at something, not being any good at it, practicing a lot, and then getting good at it. The. The other thing I think was that I sort of, like, look back on and I'm like, oh, my gosh, that was kind of an amazing experience and does have kind of, like, transferable advice, is you want to be part of a community where, like, doing the right thing is the normal thing. And for me, and like, my high school, where debate culture and competitive public speaking and drama culture was like a big, you know, social scene at the high school, doing the right thing, which was practicing a ton of. Was the normal thing. Right. If I was, like, hanging out in the debate room after school ended, you know, there would just be people off standing Facing a blank wall, delivering a speech to a blank wall, or delivering it to their friends and getting feedback from their friends, or, like, doing practice debates where they tried out different things, or, like, doing research and cutting cards. And now I look back and I'm like, okay, for. For most people who had a more ordinary, you know, like, those are pretty. Those are pretty weird, you know, habits to, like, be delivering a practice speech five days a week, you know, every week, and then also going and competing in tournaments on the weekends. It was. It was a really unusual thing, but in that community, it was the normal thing. And that is kind of like one of the lessons that I took away from debate is as you're thinking about your career, you. You want to be a part of communities where, like, the. The types of habits that lead to great performance are like, a normal, accepted part of the culture. It's part of the reason why, like, Matt, you know, I'm always asking you and everybody on the team, like, what books are you reading? Right. Because, like, it's. For me, it's. It's. It's pretty important that we are a culture of readers on this team, and I want, you know, reading a lot to be the normal thing and in our community.
- [12:37]
A
Yeah. And I can tell you that before I started here, I didn't have anyone around me constantly asking me what I'm reading. And since I started here, I've been reading much more. So it clearly works. And this reminds me of a class I took in college, which we'll get to next for you, but a class called the Game Theory of Social Behavior. And this class really drilled in on how the people around you and what they value shape your incentives. And I think that debate example is the perfect example of that, where because everyone around you valued taking a lot of time out of your day to practice debate, it became like, a socially positive thing for you to do. I think there are many communities, especially in high school, where the socially valued things to do are not nearly as helpful for your career. So I think that's really cool and a great example of how important it is to surround yourself with ambitious people.
- [13:41]
B
Yeah. And I mean, oh, my gosh, my friends from high school debate, I'm still in touch with those people today, and they're doing great stuff. Eric Min, who is my debate partner senior year, you know, now he's a professor at UCLA in political science and doing great stuff. Okay, so you mentioned college, Right. And I have to mention that you and I went to the same College.
- [14:02]
A
That's right.
- [14:03]
B
Washington University in St. Louis, which was a ton of fun. Yeah. Yeah.
- [14:11]
A
So, yeah. What did you study in college and what were some of the main experiences you had while you were there that shaped your path going forward?
- [14:18]
B
Yeah, So I was pretty omnivorous in college. I took a good amount of science classes, actually. Washu, funny enough, is, like, super strong in the evolution of humanity. So, like, a lot of big fossil discoveries have been made by Washu professors studying, like, early hominids and the ancestors of humans. Really strong at Washu. And then also primate biology is really strong at Washu. And so that's kind of where I caught the bug of AI, actually, was from the biology side of how did intelligence evolve? And I just took those classes because I thought they were completely fascinating. And I still am super interested in biology and read books every single year about biology, mostly just because I think it's incredibly fascinating. But it's also like, been actually genuinely helpful for me in my AI career that I have, like, all of this sort of reference analogies about the evolution of intelligence in not just apes, but also octopuses, and also, like, how emergent phenomena can come out of simple behaviors, like how ants are really computers running very simple programs, but if you look at the behavior of an ant colony, it's, like, super duper complicated. So that's, like, one thing I studied a lot of at Washu. The other thing was, of course, political science, because I really liked that. But school at Washu was not my main thing at Washu. My main thing at Washu was a magazine called the Washington University Political Review. And I think freshman year, I was the graphic design editor because, like, that was the only job that was open and nobody else wanted to do it. And I was like, okay, so I can, like, be on the editorial staff of this as a freshman. Great. Okay. I guess I need to learn Adobe InDesign and Adobe Photoshop and all these kind of graphic design tools. And that was, like, not anything I had any background in whatsoever. I'd never done any art in high school or earlier in life. And it was one of those instances where I was just like, okay, so like, how do you get good at something, you know, where you have no background and there's nobody who's really gonna, you know, guide you through this because I didn't have, like, amazing mentorship in graphic design. So that was just like, a lot of, like, finding graphic design blogs on the Internet and just becoming fascinated with, like, looking at examples of really good work and being like, how do they do that? And, like, Finding tutorials on the Internet for, you know, how to improve my skills, that was cool. And then when I became a junior, pretty much everybody associated with that magazine graduated. And so, like, I had to take over as editor in chief. And it was me and one other guy and a woman who was going on study abroad in the fall semester. So we literally had an effective staff of two. And we're like, okay, we have to, like, start from scratch. We have to recruit everybody. We have to do everything. And so that was like, an insane period of my life where I was like, soliciting every article. I was writing a lot of the articles. I was editing every article. I was doing the graphic design for the magazine. I was the guy carrying the boxes to physically distribute the magazines to all the bins around campus where, like, students could pick them up. And something that was like, this just crazy moment for me was like, I was so proud of myself, right. I was like, give myself a pat on the back. Like, haha, like, now we have 10 staff. I've saved the magazine. We're gonna live.
- [18:21]
A
That's no easy feat. I mean, I remember running clubs in college, and recruitment is hard. There are a lot of clubs that want your people.
- [18:28]
B
Yeah. And so I was. I was feeling great because, like, the magazine continued to exist. And for me, you know, that was success. And. And this experience that was just kind of magical in my life was I, as I said, I was like, committing my life to this magazine. And all my friends saw me committing my life to this magazine. And so they were like, oh, you're doing great, Greg. Congrats on getting another issue of the magazine out, et cetera, et cetera. And then there was this one day where I was going to the library cafe, and I saw that these two people had just sat down and they were cracking open the Washington University Political Review, the issue that I, you know, just released. And so I'm like, oh, this is so exciting. I'm gonna get to, like, actually see people, you know, read and experience a magazine. I'll kind of be a fly on the wall. And immediately these guys just start ripping the magazine apart. I don't mean physically. I just mean, like, they were criticizing every single thing of it. They're like, okay, like, why does the font, you know, randomly change sizes, you know, between this article and this other article on the next?
- [19:45]
A
Because that is a humbling experience.
- [19:47]
B
Yeah. Why. Why are there so many grammatical mistakes, you know, in these articles? And like, why is the argument in these articles bad? And it was. It was like, I was just, like, dying inside listening to these guys basically tell, you know, make fun of my baby and. And say how ugly my baby was. And so, like, you know, I'm human. So I'm going back to my dorm that day, and I'm thinking, God, these guys are a bunch of jerks. They don't even know what they're talking about. And so, like, any, you know, loser, I call my parents, you know, looking for. For reassurance, and they're like, well, you know, Greg, it is kind of weird how the font changes size between the articles. And it is kind of weird how, you know, there's a lot of grammatical mistakes and, yeah. Not all the article written all that well. And. And then I, like, went to my friends and got the exact same experience. And that's when I realized, like, everybody who cared about me in my life, my family, my friends, et cetera, they were being nice to me and telling me that I was doing a good job because they didn't want to hurt my feelings, but I actually was not doing a good job. Like, actually, the. The magazine was pretty lousy, and nobody was willing to tell me that. And I was, like, too wrapped up in how hard it was just to keep the thing alive that I didn't know that it was bad.
- [21:11]
A
Yeah.
- [21:11]
B
And so what I. What I realized is, like, initially, I hated those guys, right? I, like, I was like, I'm going to. I'm going to find some way to get back at them. But then I was like, actually, like, those are the only people who were ever honest with me. And the only reason they were honest with me is because they didn't know I was listening. Yeah. And so then I, like, you know, reopen the magazine, and I'm, like, looking. Looking at it with fresh eyes. I'm not looking at it as, like, my baby, who I love. I'm looking at it like, okay, what is everything in this magazine that is not awesome? And what does it take to. To get from, you know, finished but lousy to finished? And it's also awesome. And that. That, for me, was, like, one of the defining moments of my whole life because it, you know, taught me the gift of criticism, right? Like, people who tell you white lies, they're usually doing themselves a favor, right? Like, because they don't want to be honest with you because it would be too painful for them to, you know, give you the honest feedback of, like, this is not going well and you're not doing a good job. And so anytime you can, like, get somebody who actually knows what they're talking about. Who is willing to, like, point out everything that you're doing wrong. That's, like an incredible gift, you know, that's, that's like, so amazing. And so I really shifted my mindset in terms of, like, looking for feedback to like, hey, what can I do? What can I do better, right? Like, oh, you know, maybe they'll, they'll, they'll say like, oh, great job, Greg, etc. Etc. It's like, well, thank you. But that actually doesn't help me that much. Like, you know what really helped me is if you tell me, like, what I can do better. And that was like, my number one takeaway from the Washu Political Review is just being willing to seek out constructive criticism. And even as it's difficult to give constructive criticism, it's. It's actually one of the nicest things you can do to people.
- [23:11]
A
Yeah, that's great. And I'm really glad at the end of the day that you kept the Washu Political Review around. I mean, it's still around today and you wrote for it.
- [23:19]
B
Right? Which is. That's right.
- [23:20]
A
And, and funnily enough, my writing sample for all the jobs I applied to was my piece that I wrote in the Washu Political Review. And what I wrote about was also your area of expertise, which is expert controls. And so I wrote in your magazine on your area of expertise, and here I am today. So that is a weird coincidence, but also a happy one.
- [23:42]
B
Yeah, a weird coincidence, but also becoming less and less shocking that you work here now. Right?
- [23:47]
A
Yeah, that makes sense. So I want to move on to after college. Your first job out of college, I believe, was at a consulting firm. Can you tell me a little bit about that?
- [23:56]
B
Yeah. So I worked at a boutique consulting firm called Avicent. It has since gone away. It was acquired by a different consulting firm. When I got there, it was about 60 people and they were in the midst of a big growth phase. So if you, you know, if you have done or know people who've done the McKinsey B. BCG type consulting, it's that type of consulting. So it's not like.
- [24:26]
A
So management consultant.
- [24:28]
B
Yeah, it's management consulting. Like corporate strategy type consulting was the. Was the main focus over there. It was not like a politics, lobbying, esque kind of consulting firm. And this was like another huge wake up call for me because, you know, like, you know why professors read everything you write? Because they get paid to read what you write. And when you are in, like, the working world, you have to actually produce work that is valuable to the reader. Otherwise people won't read it or you'll get fired. And the thing about avocent that was magical. I mean, there was a lot of things that were pretty magical about avocent. It had a really lovely corporate culture, really lovely culture of mentorship and other stuff. But one of the more remarkable things was that they had an ethos where the junior analysts, if you had done the research, you got to be in the meeting when we briefed it to the executives of the company or government agency. I did a lot of work for NASA, did a lot of work for the Air Force on space and satellite policy issues, among other things. And if you did great work, you got to be in the room. And if you did really great work, you got to brief your part of report, which in most cases is like PowerPoint slides. And so the crazy, you know, miraculous aha moment that happened for me in my first year at Avison was we were working for this satellite company. They operated satellites in space and they made money, you know, off the communications that you. Communications capacity that those satellites provided. And they had, you know, a contract with the Department of Defense, basically selling satellite communications for people who were fighting in the wars in Iraq and Afghanistan at the time. And their big contract, you know, was up for renewal. And they were worried, you know, that they weren't going to get the contract renewal. And so they asked us, you know, help them with their corporate strategy to, you know, get the contract renewed. And so I'm like, satellites, you know, like, I liked space. I was really interested in it, but I was like, by no means an expert. And so, like, the idea that I'm going to have to, like, advise a, you know, CEO of a billion dollar corporation, you know, on, like, their satellite strategy was like, on the one hand, like, exciting, but on the other hand, terrifying. And that was, you know, I sort of like leaned back on my debate career because I was like, okay, how do you, how do you win an argument like, how do you do anything? And it's like, okay, well, I just graduated from college like a year ago, so I have no credibility whatsoever. So, like, the only way that I'm going to be useful is if I show up with data or if I show up with like, other authoritative sources that are useful. And I put that together in a way that is useful and meaningful, but does not rely upon the gravitas of me because I don't have any gravitas. I just got out of undergrad. And so in that case of that company, I managed to find this alternative program that was going on in the DOD where they were going to build their own satellite. And the satellite that they were building was specifically targeted at, like, many of the features that this company thought that they were unique in providing as a commercial provider of satellite communications capacity. And the other thing I did is that, like, I got a bunch of data about, like, when they sold the little handsets through which, you know, people could. Could talk on their satellites. And I noticed that, like, literally, sales of their handsets. I, like, mapped the data because. And I saw, like, this big arc. And I was like, you know what? That ARC looks a lot like troop deployments to Iraq and Afghanistan. And ran a linear regression. I'm like, literally, There is a 98% correlation between when troops get shipped to Iraq and Afghanistan and new detachments and how many, you know, handsets you're selling. And so I, like, my boss loved my analysis. And so I got to brief it to the CEO, right? Like, this is the CEO of a company worth billions of dollars. I'm like, I think I was 22 at the time. And so I'm like, yeah, so your contract is up for renewal, but here's this military program that's going to build a satellite. And they say that the military satellite that they're going to own is going to do everything that your satellites do that you rent to the government. And then also, here's this chart that shows that there's a 98% correlation between sales of your handsets and troop deployments in Iraq and Afghanistan. And I'll never forget this moment. It was, like, one of the most amazing moments of my life. The CEO turns to the. I think he was like the executive vice president of government sales or something. He turns to the government CEO and he's like, why am I learning this from him? And he points to me, right? And the CEO is finding out the stuff that I'm telling him for the first time. It's blowing his mind. And he, like, can't believe that, you know, I'm the one telling it to him and not, like, the people who work for him. And, I mean, it was a bad day for that guy, but it was a really cool day for me because I realized, like, okay, wait. If I really put in the work, if I really, like, research everything, you know, at that time, it meant not just reading a bunch of stuff, but, like, actually picking up the phone and calling people in the government to, like, ask about this other military program or to other stuff. Like, if I actually put in the work, I can do something that is a meaningful intellectual contribution that people who matter will actually find useful. And, and that was just so exciting for me. I was like totally hooked, you know, based on after that moment. And Avocent was like an amazing place for like giving me those kinds of opportunities. And for me it was really about like, like how to do research in a way that doesn't rely upon my personal credibility, which I don't have, but is nevertheless like meaningful and impactful and like powerful. People will change their minds based on it.
- [31:07]
A
Yeah, yeah, that's a great experience to share and I want to, I want to get to our next section on just general career advice, but before we get there, a few more experiences that I want to hear you talk about. Yeah, your time in graduate school and then your two jobs that you had before you came to csis, which were at Blue Origin first and then the DOD Joint AI Center.
- [31:30]
B
Yeah. So I think a lot of graduate school will probably talk about in the question of like, should I go to graduate school? But let me just say, like, this was a wonderful time for me. But like, for me, a lot of the value of graduate school was the internships. So I did a three year program, the joint MPP MBA at Harvard Kennedy School and Harvard Business School. So two summers to get internships. But I also had internships like during grad school, like during class time. And I worked at iRobot, which makes the Roomba Manu. The Roomba, like vacuum cleaner is one of their big claims to fame. And so they were like a big robotics company in Boston. And I got to work in their corporate strategy unit. I worked for a robotics startup that was using AI for like grasping stuff. I got to work for Samsung in their corporate strategy unit in Seoul, Korea, on AI optimized semiconductors, sort of corporate strategy type tasks. And then I got to work at the White House Office of Science and Technology Policy, you know, as a measly intern working on space and satellite stuff. So like, and for me that was awesome. And I think my big takeaway from graduate school is there's what you learn in class, which is super valuable. But at least in the case of my university, Harvard Kennedy School, there's people who were super powerful fancy people not that long ago who are now teaching there. And so I definitely like had like a laser focus on, you know, I want to come away with a relationship with Joseph Nye who, you know, coined the term soft power and ended up being my thesis advisor.
- [33:22]
A
Yeah, sounds Like, a good person to know.
- [33:23]
B
Yeah, exactly. And, like, I wanted to have a relationship with Megan o', Sullivan, who had, you know, been really influential on the Bush National Security Council. And so I was like, like, okay, I not only want to, like, take these classes, like, these are the classes that I need to stand out in, because I don't want. I don't want to. I don't really even care about getting an A. I care about, like, learning, and I also care about, like, impressing this professor enough that they'd actually be willing to, like, make some introductions for me and help me in my career. And so, like, that was sort of the criteria I was optimizing for in graduate school. And it all worked out pretty well.
- [34:00]
A
Yeah. And then next up, we had Blue Origin.
- [34:02]
B
Yeah. So Blue Origin is a space company founded by Jeff Bezos, also the CEO of Amazon. This was my first job for two years after the end of graduate school. And among other things, like, I had to write the sort of, like, weekly corporate strategy memo that went straight to Jeff Bezos. So that's like 2,000 words every week on top of, like, reports that I have to write. So there's like, the weekly update thing, and then there's also, like, the sort of, like, standalone reports that I had to write on any given topic. And they're going to Jeff Bezos, they're going to the CEO of the company, they're going to every leader on the team. And that was like the weekly heart attack, you know, hitting send on an email to Jeff Bezos when you're one year out of graduate school. Pretty dang stressful. But, like, that was, for me, like, when my writing really, really took off and, like, when I became what I now consider myself to be, which is a good writer. Yeah. And the. The reason why that happened is because I got in a lot of reps. You know, I had to. I had to shell out 2,000 words every week. And, like, was I trying to make those 2,000 words awesome? Yeah, it was going to Jeff Bezos and, like, every executive at the company. And so this. The sort of lesson here, right, is that you want to find a job that allows you to get better at the skills that you want to get better at and to do a huge volume of work in those skills. So, like, on the one hand, like, having that weekly memo to write, having, like, the, you know, multiple reports in multiple, like, you know, 50, 100 page reports I had to write a year was like a heart attack inducer, but it was also, like, forced practice at an incredibly high volume with real stakes.
- [36:09]
A
Yeah. And the fact that the stakes were so high and you're sending an email to Jeff Bezos makes you want to make sure that that email and the briefing is really high quality.
- [36:18]
B
Yeah, no, like that was it, like it was, it was, it was kind of this amazing forced practice situation and worked out really well for me.
- [36:29]
A
Yeah. And then last but not least, before csis, tell me a little bit about your time at the dod.
- [36:34]
B
Yeah, so the, the way I ended up at the DOD was like I had written my master's thesis in graduate school on artificial intelligence national security, and that work was getting a lot of media attention because the Harvard Belfer center published it. Even though it had been my master's thesis, they just published it like it was a regular think tank report and it got written about and Wired and a few other media publications. And I'd already accepted my job offer at Blue Origin. And so I was like, well, gosh, I don't want to give up this AI think tank work because I'm really finding it very rewarding. But it can't be my job because my job is going to be corporate strategy at Blue Origin. And so I took that thesis after it was published and I went down and I met with some think tank folks here in D.C. and I'm like, hey, I wrote this thing, can I be a non resident fellow in your program? So my job application wasn't like submitting a job application. My job application was like, hey, I'm the guy who wrote this thing is good. If you let me be a part of your think tank, I will write more things that are also hopefully going to be good. And that was like, turned out to be a great tip that I now give people which is like, don't ask for the job you want. Show that you can do the job that you want. Have a body of work that demonstrates you have those skills. So Blue Origin was my day job center for New American Security where I was a non resident fellow, was like my nights, weekends and vacation days kind of a job. And when my work started getting enough attention, including of like Lt. Gen. Jack Shanahan, who was a three star general in the Air Force, leading a bunch of really important DoD AI projects. You know, when he was setting up this thing called the Joint AI center, he said, hey, would you come on and be our chief of strategy? Which was an incredible opportunity for me. And when I got to the Jake, you know, I encountered all of the difficult things that you encounter when you join the Government, like I think it took three weeks before I got a computer, you know, so like literally they didn't have a computer waiting for me. And there's just a lot of ways in which the government is like this incredibly high achieving, very inspirational organization. And then there's these other ways in which it's like dreadfully backward and incompetent. And what's weird is that both of those things are true. Like they exist side by side. I think one of my like big takeaways from my time in the Jake, other than just like, how does the government work? How do bureaucracies actually carry out policy stuff that like, there's plenty of things that I said when I was a non resident think tank fellow and I was like, like, you know, the government should do this, America should do that. And like now I never write those kind of recommendations because I've been in the government and I'm like, yeah, those, those are not useful, not happening. If you can't name the office or the individual who is actually going to be responsible for doing the thing that you are recommending get done, then assume your recommendation is not going to get done. Right. Like, so what I really, what I really honed in on when I was a government, you know, policymaker was like, you want to have yesable recommendations. Like there is somebody in government who could conceivably say yes to your recommendation and actually has the authority and the capacity to carry it out. And the second big takeaway was like the difference between those two things, the difference between authority and capacity, like what the government says it's going to do versus what it actually does and can pull off. There's oftentimes a huge delta between those two things. And I learned those lessons very painfully and viscerally during my time in the government. Even though I had a lot of wonderful experiences while I was in the Department of Defense. I think the fact that government is a thing made out of human beings and organizations that, you know, are fallible and can be dysfunctional and thinking about how do you achieve policy success, even though you know that is true and not getting, you know, just completely discouraged and thinking there's nothing good to be done, that was like my, my, my big takeaways from the time at the DOD.
- [41:14]
A
Nice. And, and that brings us to CSIS, where you've been since 2022. We're probably going to talk about a lot of your takeaways from working in AI policy at csis. In the last section, the AI policy section. But briefly, what are some of the big takeaways you've had.
- [41:33]
B
I think I have had a pretty good string of luck at CSIS in terms of writing stuff and getting the attention of policymakers and having policies that I recommend actually become American policy sometimes because I was saying something that other people were also saying, but sometimes saying something like that really did come from me, and I was the originator of the idea, and that's really exciting. And there's a lot of different ways to, like, be successful at think tanks. My way is not the only way. And there are other people at CSIS who have, like, a very different approach to being successful at think tanks. But the way I really think about it is it's very much about, like, the written word and the publications. And what I am trying to do goes back to my time in consulting and goes back to my time at Blue Origin where, you know, I'm trying to comment on an issue. And now that I'm at csis, I mean, CSIS is like a very good, good brand. You know, back when there were think tank rankings put out by, like, the University of Pennsylvania, CSIS was routinely named the best defense and national security think tank in the world and a very, very good think tank on a bunch of other rankings. So, like, you know, there's the CSIS brand, But when I'm writing my think tank reports, I, like, completely ignore the think tank brand. I completely ignore the CSIS gravitas, and I go right back to, like, my time as a consultant and right back to my time as Blue Origin. And I'm like, why would anybody care what I think? You know, I'm nobody. And so I always try to, like, show all of my work in. In any argument I make, right? Like, if you. If you have ever read any of my reports, you will notice that there is, like, a preposterous number of hyperlinks in them. And that is because. And even when there's not a hyperlink, there's usually, like, an explanatory warrant. And that is because I try to never have a fact in my think in my reports where, like, the. The source of my claim is not transparent to the reader. And I never try and make an argument without, like, giving you the supporting structure of here's why I think that. And, and the, the way that I think it's worked out for me, when it works out well, is that that's a. That's a style of writing that is very easy to trust. And it's also very easy to understand, right? Like, whether you're start. Whether you're reading my stuff from, like, a very high level of expertise or, like, a very low level of expertise, I'm going to walk you through every single thing that I think and why I think it in the clearest, simplest language I can possibly use to describe it. Because my assumption is, you know, you as the reader, don't trust me at all, and that I want you to end up agreeing with me even though you don't trust me, because I've been so transparent about, like, every step of my argument that, you know, what else is there to disagree with? Right, Right.
- [45:03]
A
Yeah. Well, it's been fascinating hearing more about these experiences, and I've heard a few through lines throughout all these experiences that I want you to sort of touch on in this next section on general career advice. So the first one is just going to be the importance of doing really great work. This is something that you and I have talked about before, and I think it would be good if you just started out by explaining what is it that you mean when you say great work and what does it take to consistently produce it?
- [45:36]
B
A fabulous, fabulous question. Like, it really comes back to, you know, a few things, right? Like I said, you know, when you're applying for a job, you kind of want to be able to show them that you can do the job. You want to kind of, like, have the proof. And another way of, like, framing this is, you know, in the think tank world, there's such, like, a fight for relevance, right? You, like, you want to be the person who policymakers, you know, care what they think. And I always say this is, like, this is a bad sign if you're focusing on this, for sure, the first thing to focus on is to deserve to be taken seriously, right? Like, the first thing is not to be taken seriously. The first thing is to deserve to be taken seriously. So, like, how do you know what you're talking about? How do you learn to write in a way where you can communicate your ideas in a way that is understandable and useful and recommendable? And there's, like, no substitute for practice, you know, like I said in my debate days, right? Like, hours and hours every, every day, every week, I was, like, practicing this stuff to get better. And when I was working at Blue Origin, I was, like, writing and, like, thinking, like, constantly, like, how can this get better? And so there's, like, a few things that I would say here is, in general, I just don't believe there is such a thing as a natural at any skill in the world. I Mean, like, obviously people had genetic predispositions. Like, Michael Phelps's feet are practically flippers, so he's really good at swimming, you know. But in, in almost all skills, I really think that if you are good at practicing and if you are good at specifically doing a sub discipline of practice called deliberate practice, which is the term, if you haven't heard it, I strongly encourage you to Google it. It's really about, like, knowing what you're trying to get better at, carving out time in your schedule to practice it. And as you're practicing, focusing on what you're not doing well and what you're trying to change to get better at it, rather than coming back again and again and again to like, what you're already good at and enjoying that you're already good at it. Like, that's not going to help you get better. What's going to help you get better is to like, identify where your weaknesses are and working on improving them them. The other thing is to, like, study great work. One of the habits that I picked up from the graphic design days at the magazine, which, like, I'm still like, not a champion level graphic designer by any means, but I'm like, way better than I was. And part of the reason why I am is I would just like, look at people's amazing work and I'd be like, like, how are they doing this? Like, why is this so good? What did they do that made it so good? And so like, in, in so many aspects of my career, I don't just like, admire great work. I like, jealously admire great work. And I'm like, how did they do that? Could I do that? What would I have to do to do the thing that they just did that I thought was so impressive? So like, when I read, you know, great writing, I will like, take time and just be like, okay, why is this so effective? Like, why did this intro work? Why did this, you know, reframing of an issue, like, resonate with me in a way that it didn't before? When I see like a great public speaker, I'm like, why are they a great public speaker? Like, what are they doing that is effective? And then I'm like, okay, can I basically copy that? Can I figure out how to be good at this thing that they just proved is a thing, that it matters when you're good at it? And then it comes back to like, seeking out criticism and basically, you know, not going to your mom and dad and being like, didn't I do a great job? But like, Going to people who you know are awesome and produce great work and asking them to rip your stuff apart. That's how you get better, is having your stuff ripped apart. And when I was in consulting, I had, you know, a wonderful boss, multiple wonderful bosses. But, you know, this one guy named Royce Dalby, who was a really important mentor in my life, and he would, you know, just basically, like, wrong, wrong, wrong, wrong, wrong, wrong, wrong. You know, bad, bad, bad, bad, bad, bad, bad. Great job. Like, this is going great. And it was just the most loving. This, like, it was just this loving way of tearing me apart and building me back up at the same time. And I got so much better so fast. You know, working. Working with him. It was, yeah, like an amazing time in life. So if you can find that. If you can find people who are good at the things you want to be good at and are willing to give you constructive criticism, that is just, like, a gift from the gods. So study great work. Build time in your schedule to practice and, like, find talented people who can give you harsh but true feedback and, like, be willing to take it. I think those are, like, critical foundations of doing good work.
- [51:17]
A
Yeah. And I think something important here is that it might be easy to be discouraged if you're, like, early in your career and you're producing work and you don't feel like it's great work. Right. But the fact that you can tell that the work you're producing isn't great is actually a really good sign.
- [51:32]
B
Yeah. And this gets to, like, something that I have always loved. It's a quote that is just like an ex. An excerpt from a random interview that Ira Glass gave at one point. And it's something I've come back to again many, many, many times in my life. And so I'm just going to read it because I think it's so useful. For those of you who don't know, Ira Glass is the host of the podcast and radio show this American Life, which created many of the conventions of the modern podcast even before podcasts existed. So this guy's, like, a legend in the public radio world. So here's the quote. Nobody tells this to people who are beginners. I wish someone told me. All of us who do creative work, we get into it because we have good taste. But there is this gap for the first couple years, you make stuff, it's just not that good. It's trying to be good. It has potential, but it's not. But your taste, the thing that got you into the game, is still killer. And Your taste is why your work disappoints you. A lot of people never get past this phase. They quit. Most people I know who do interesting creative work went through years of this. We know our work doesn't have this special thing that we want it to have. We all go through this. And if you are just starting out or if you are still in this phase, you gotta know it's normal. And the most important thing you can do is do a lot of work. Put yourself on a deadline so that every week you will finish one story. It is only by going through a volume of work that you will close that gap, and your work will be as good as your ambitions. I think that, like, just 10 out of 10 on that quote is like, I stunk when I started. I stunk at public speaking before I did debate. I stuck at writing before I did the Washington University Political Review, and then again in a different way, stinking at writing. Before my consulting job and before my Blue Origin job. And you just have to, like, refuse to be discouraged because the, you know, everybody who's ever been good at anything started out being bad at it.
- [53:55]
A
Yeah.
- [53:55]
B
And you just have to carve out time for the deliberate practice and do it. So, like, whatever it is that you want to be good at in your life, take a look at your calendar and, like, ask yourself, when am I, you know, making time for this? I mean, one of the things is, like, you know, just, like, subject matter knowledge and, like, things you can count, I think. I think things that you can count are, like, really helpful, even when they're misleading. So, like, I try to, you know, I track the number of books I read every year, and I try to always hit at least 26. So, like, once every two weeks, and on a good year, you know, I might hit 52. And it's not that there's a direct correlation between I've read more books and I'm smarter. And not every book I read is some AI policy tome or some biology manual. But what really matters is the habit of reading. And so by counting that, I know that I'm carving out time to always be learning, to always be learning about China, which I'm so interested in. To always be learning about AI, which I'm so interested in. To always be learning about history and the history of organizations, both companies and governments. And I just like, when I talk to people and we said this at the beginning of the podcast, it's like, okay, you want to be an expert, but you're not reading books. Something wrong here? This is not going to go well for you. And it's true for any skill you want to learn, whether that's an out and out skill, like public speaking or writing. Like, the acquisition of subject matter expertise is, you know, are you looking at it? Are you identifying your weaknesses? Are you practicing to try and round out those weaknesses?
- [55:50]
A
Yeah, that's great. And another through line I heard throughout, a lot of the anecdotes you told in the, in the last section was how important it is to put forward clear and good writing. I think you have a lot to say about this topic, so I just want to hand it over to you and talk about how can you get better at writing.
- [56:09]
B
Yeah. So, I mean, I think I would separate this into like, hacks, like tricks, and then like, like the, the longer term, like, how to get better at practicing.
- [56:22]
A
Sure.
- [56:23]
B
Let me start with like, the hacks. I, I am, I'm doing myself a disservice, you know, because I'm creating a better generation of competitors in the think tank game by sharing this hack.
- [56:36]
A
We appreciate your generosity, but, like, I.
- [56:39]
B
Cannot believe more people don't do this. Read your work out loud. Literally print it out on paper. Put your phone in a drawer and get a red pen and read your draft out loud. There are so many things about good writing that your ear understands and your eyes don't. And I don't know what it is about the human brain that makes that true, but something about reading your work out loud will just make manifest to you all of the ways in which your writing is currently clunky or confusing or drones on, etc. Something about hearing it really puts you in a position that is closer to the reader because as a writer, you are a terrible reader of your own work. And the reason is, you know what you mean, you know what you're trying to say, but the reader can only know what you're trying to say if the writing gives it to them. And by reading your work out loud, somehow you just become closer to embodying the Persona of the reader and you can identify all of these weaknesses. I would say, like, literally no matter where you are as a writer in your life right now, whether you are garbage or gold, if you just start reading your work out loud, it's like immediately plus 25% to your writing skill. So that's like, for sure the number one hack. The second thing is to argue against yourself, right? So like to be the editor of your own writing, but to do so with like a critical lens. So I, I mentioned you Know how as a debater and how I had to argue both sides of an issue. And so I kind of like put on my adversarial reader Persona when I read my work and I'm like, this sentence, how could it be wrong if it was wrong, how would I argue that this sentence is wrong and you know, this paragraph, if it was wrong, how would I argue it's wrong? And so I try and you know, not just cite every single fact that I'm including, but also like walk you through the logic so that it's very hard to disagree with me. Especially if you're doing so in like a sincere way and not like a, you know, maliciously motivated way.
- [59:20]
A
Strong and.
- [59:20]
B
Yeah, yeah. And I don't mean that like, you know, I'm always right and that everything I've written is gold, but I just mean that like adopting this kind of Persona of if I was going to tear my work apart, how would I do it? And how can I, how can I then rewrite my work? To anticipate those objections and rebut them in advance is just a lovely kind of a habit.
- [59:44]
A
It nice.
- [59:46]
B
And then the final thing I will say is seek out criticism of your drafts. And I mean this from everyone, right? Like it does not matter if you are William Shakespeare or Ernest Hemingway. You need other people to read your drafts like always. And they don't just need to be like experts or like geniuses. Like don't get me wrong, you know, we wrote a paper on Middle East AI last year. It's awesome that we have somebody at CSIS like John Alterman who's like a legend in Middle Eastern policy studies, you know, to do a prior review of that. But you know what else helped was like the review of our interns. You know, like I actually love it when non experts are willing to do a read through of my work and I like literally tell them like, like write a C next to any paragraph where you got confused. Write a B next to any paragraph where you got bored, right? Because like that's what I want to understand because like I know what's in my head that I'm trying to communicate when I'm writing and I am lying to myself that I am effectively, you know, communicating it all because I'm bringing the baggage of my own thoughts to it as I read it. So like a fresh pair of eyes is awesome and the diversity of backgrounds is like even more awesome because you just never know what you don't know about how people are going to read your work. And how it's going to confuse them or excite them or anger them. And as much feedback as you can get is awesome. And again, sorry, as you're consuming all this feedback, you got to have one mindset. Criticism is a gift. Criticism is a gift. So you got to ask these people. It would help me a lot if you would tell me where this is boring. It would help me a lot if you tell me where this is confusing or where you think it's.
- [61:39]
A
Yeah. And this brings me to a next topic that I want to talk about, which is mentorship. I think it sounded like you had some great mentors throughout the course of your career. How important do you think mentorship is for succeeding?
- [61:50]
B
Oh, my gosh. Like, so many people have, like, meaningfully changed the course of my life. My debate coach in high school, Chris Riffer, totally changed the course of my life. Professor Andrew Rayfeld at Washu changed the course of my life in political science and other stuff. Roy Stalby, I mentioned at avocent, changed the course of my life. And Joseph Nye changed the course of my life at Harvard. And it's. And, you know, even now, I still have mentors, you know, folks like Richard Danzig who have just been so generous with their time to me in a way that I. I never really understand why they're willing to take so much time to teach me. Oh, Jeff Seglin, there's another one. I should say. He was at graduate school. He was my writing professor. And you really, really, really, in the same way that you want to be part of a community where doing excellent work is, like, the normal thing, practicing on skills is the normal thing. You also want to find a community where, like, mentorship is. Is a part of the game. And here's the thing. Mentorship is, like, very, very expensive. Like, it takes a lot of time to teach people, especially in a. In an industry like the think tank world, which is very artisanal and kind of apprenticeship style instruction. And so finding somebody to take you seriously and mentor you is amazing. The best advice I can give for how to find great mentors is to ask yourself, what would it take to be worthy of a great mentor? Like I said in college and in high school, people are paid to be your mentor. They're paid to help you out with stuff. Stuff. But in the real world, you know, they're not going to be paid to be your mentor, presumably. And so you have to say, like, I want to be worthy of their mentorship. And I have asked people to, like, go through this thought Experiment, you know, at various times, like, I have to write recommendation letters, right? People come work for me and then they want to go off to graduate school, they want to go off to another job, etc. And one thing I've said at various times is like, all right, you know, here we are, you're working at csis in a year or two, you're going to ask me for a recommendation letter probably, right? Maybe you'll totally blow it and you'll know it and you won't ask me for a recommendation letter, but presumably you're going to ask me for a recommendation letter. Why don't you write that recommendation letter right now? Like, we haven't, we haven't worked together at all. Go write the recommendation letter that I'm going to write for you, right? About how tirelessly, how tirelessly hard working you are, about how innovative you were in solving problems, about what a good sport you were when times got tough. Like, talk about the person who would be the perfect recommendation letter and then go be that person for the next two years, you know, then go make it true. I think it's just like a nice little way to think about it. So, like, when you think about being a mentee and seeking out a mentor, like, ask yourself, like, what would the perfect mentee look like for this person? And then go try to be that person so that they will want you to be their mentee. Wow.
- [65:38]
A
Yeah, I think that's really good advice. Maybe I should spend my afternoon doing that.
- [65:42]
B
Yeah.
- [65:43]
A
But last thing I want to talk about, in the umbrella of sort of general career advice and doing good work, I think it's easy and also always true that you have a lot of uncertainty when you're making decisions about your career. One piece of advice that I've sometimes received is it's really important to experiment. And by that I mean taking low cost at first, ways to learn whether you're good at something, whether you like something, and then as you learn more about yourself, taking higher and higher cost opportunities that maybe take more time, maybe take more effort, will give you more insight into how good of a fit you are for a certain role or a certain career. I think on the flip side of that, it can be really easy to sort of drift. And by that I mean find yourself in a position where you're comfortable or take the path of least resistance. So I'm curious what, what role experimentation played for you and how important you think it is generally.
- [66:43]
B
I mean, it's funny you say this, but like, my own perception of My own career is just like crazy stroke of luck followed by crazy stroke of luck. So I graduated in 2010, which the financial crisis and the Great recession happened in 2008, but the actual peak for youth unemployment happened in 2010. So I graduated into a rock bottom, you know, job market, the worst in many decades. And man, I got that job at Avocent. And let me tell you, I did not have great other options. You know, like, I probably would have been able to earn a salary, but, but would I have? I had nothing else that was even remotely that good. And what's funny is I didn't even know how good I had it. I was like, oh, abacy, you know, consulting, it seems fine, but now as I look back, and I was like thinking about like, what made it amazing was just how much like skill acquisition I was able to get, you know, while I was there. You know, how to do research, how to brief research to senior officials, how to write a report, how to like, learning all this subject matter expertise and you know, not just the space industry, but the high performance computing industry and like how to switch between topics and then, you know, get up to speed fast in a way that you can be ready to brief the CEO in like eight weeks, you know, starting from potentially a very low baseline. And so if I was to, you know, give advice as to like, what, what people should want out of their, their early jobs, I do think that you really should be optimizing for learning. Like that is the number one criteria for your first few jobs. Like, it would be dreadful if you started a job and you got that job because you were good at something, you did the thing you were good at for two years and then you leave and you're still roughly as good as you were when you started. That's like a tragic, tragic outcome. Yeah, right. Like an awesome outcome is you start a job and you're good at some things and you're terrible at some things and you leave the job and you're great at some things and you're good at a lot more things. And here's the thing, like learning, learning matters so much more to young people than it does to old people. I mean, learning like is like an obsession for me even now that I'm, you know, a middle aged father of three with a mortgage and stuff. Stuff. But, but like if you just graduated from undergrad, any skill you learn, any subject matter expertise that you acquire, you get to use it for 40 years. If I learn something, I only get to use it for 20 years, right? Assuming that we all retire at 65. But, and so I think optimizing for learning is worthwhile. And, and in terms of where to go, I do think that it's worth asking yourself, if I do amazing work, how am I going to convert that into amazing opportunities? So I said Avocent was a phenomenal learning opportunity for me. But it's also consulting where all the clients are confidential. You can't like talk about your work. Anything that you do that's really meaningful, the client gets all the credit. You know, you sort of stay in the background. And so I was, I was in this situation after like my fourth year at Avocent where I was like, well, I feel like I'm doing super, super cool stuff, but I have no idea how I'm going to like get from where I want to be to like the jobs that my heroes have and that I like dream of, you know, being at some point, point. And so that's why it was useful for me to go to graduate school where I basically, I had a resume that I thought had a lot of cool experiences and accomplishments on it. But Avocent was a boutique consulting firm that most people hadn't heard of. And so I wanted to convert that into a credential that was more widely accepted and more universally understood. And then at the same time I also wanted to, you know, learn new skills, meet people in circles that I was trying to break into, et cetera, et cetera. And so that's, that's kind of how I think about early career. I mean, of course there's money, you know, which is a part of this. It, you know, like sometimes good opportunities don't pay well. But what I always say is like, like, you know, ideally you should be getting paid well in dollars and in learning, but it is unacceptable, right, if you're not being paid in learning.
- [71:52]
A
Yeah. Well, I think this is a good point to go into our AI policy specific career advice. So we've been talking a lot about some sort of high level ideas that are important for success generally. But let's talk about lessons that will help you succeed specifically in the field of AI policy. I think a good place to start for this is just framing this by how you think about what it means to succeed in AI policy.
- [72:22]
B
Well, I think for us success is policy impact. And the way that I often talk about think tanks is imagine a spectrum, spectrum. And on the left side of the spectrum is the most corrupt Machiavellian lobbying firm you've ever heard of. All they care about is Winning. They don't care about the truth, they don't care about scholarship. All they care about is, you know, changing the policy to be what they want.
- [72:54]
A
There's some of those out there, right.
- [72:55]
B
They certainly exist. Right. And then on the, on the right side of the spectrum, imagine the most, like, ivory tower university scholar you've ever heard of. Like, all this person cares about is the pursuit of truth. It doesn't matter if nobody ever reads their work. Maybe they don't even publish their work. Right. Like, all they care about is the pursuit of knowledge and truth. For me, like, a good think tank is in the middle.
- [73:17]
A
Yeah.
- [73:18]
B
Which is to say we care about scholarly rigor, we care about advancing knowledge, we care about the truth, but we also care about policy impact. And I want to do things to be effective. And so for me, like success in a think tank and in AI policy is you're actually advancing the state of the knowledge in the field. You know, you are figuring things out that matter when they are figured out. You are bringing ideas to the table that could be done. And then, you know, if you're lucky, nobody bats a thousand. But if you're lucky, then, like, those ideas become real. People actually go do those things. And I've been fortunate enough to have that like, a bunch times in my career. And I will say, never get sold. It's awesome. Every time. It's so cool when you can, when you can write stuff down, throw it out on the Internet and then see the little ripples, you know, changing in the world.
- [74:13]
A
Yeah. I want to talk about the 8020 rule or the Pareto principle and hear how you think it applies to AI policy. For those who aren't familiar, this is the idea, at least in the context of careers, that about 80% of the impact in any given field comes from 20% of the experts in that field. I think maybe the easiest way to understand this is that around 80% of the citations in any academic field come from the top 20% of researchers. So do you feel like this is the case for AI policy?
- [74:45]
B
Yeah, definitely. And I would say it's a broader phenomenon. I mean, the Prato principle goes well beyond think tank world. When I was in undergraduate economics class, we were taught this phenomenon called superstar economics, which is that a disproportionate share of the returns in some economic field accrue to the top performers in that field. So one way to think about this is the best musician in the world in the year 1800 was making how much more money than a good music like an average good musician in the year 1900. Right? So like maybe the best musician is like performing at the Vienna Opera and then like a good musician is like performing at like a saloon in New York. Well, you know, in the olden times when you know, the, the maximum number of people you could perform in front of was like in the, you know, probably single digit hundreds and a nice concert hall, the best musicians in the world probably made something like 10 times, 50 times more than an average musician. And then fast forward to like the 20th century when like vinyl records are invented or. Well, I guess they were invented before that. But like before the, you know, modern music industry takes off and the radio takes off and the Beatles are earning something like 10% of all the money in the global music industry. Right? Like just crazy outsized retailers returns to being on top of a given field. So you see this in music, you see this in sports, you see this in like a lot of fields. And I would say like it holds true even at like microscopic niches, you know, of like it holds true in think tank world. It also holds true in the subfield again, you know, of AI policy, think tank world. And so that's why I'm always obsessed, you know, with making my papers as good as I can possibly make them. Because my mental model is because like when you're a policymaker, when you're in government, it is just a fire drill every single day, right. So your dream, right, is that you've read everything ever written on a given topic, but like your reality is, you know, you're just trying to stay above water with, with all the crazy stuff that's coming at you every single day. So my, my mental model for think tank papers is that like a good policymaker is going to read the best paper on a topic, the first good paper on a topic, and probably nothing else. Right. So either, either you win because you're the fastest, like you have the first good paper on a topic, or you win because you're the best. Yeah. Or you're probably not getting read. First approximation. I mean, I'm definitely oversimplifying here in telling this story. And it's also true that like, like within ideological factions, the story is a little bit different. Right? So like, you know, the first good libertarian take on an issue can go somewhere. The first good MAGA take on an issue can go somewhere. The first good, you know, progressive take on something can go somewhere. But to a first approximation, quality really, really matters.
- [78:09]
A
Yeah.
- [78:10]
B
And so I'm always like obsessed with, you know, trying to make my, my papers as good as I can possibly make them. Because my, my sense is that the returns are so skewed towards the top. You know, that the difference between going from 10th best paper to 9th best paper is like going from irrelevant to irrelevant. And the difference between going from, you know, third best paper to second best paper is like maybe a ton of readership.
- [78:40]
A
Yeah.
- [78:41]
B
And then going from second to first is like a ton more readership. So for me, like, the obsession with quality has been sort of like a good guiding light. And how do you have quality? By ripping your work apart and trying to find every single thing that's wrong with it and by going and asking people what's bad about this. And, you know, eventually, once you don't have any imperfections, you'll be perfect. And like, in reality, nobody ever reaches perfection, certainly not me. But I still think that's like a good, good mental habit to have in trying to do work.
- [79:13]
A
Yeah, well, you started hinting at this a little bit, but how important is specialization then? Because it feels to me like if you want to be in that 20% and you want to have the best paper on any given topic, well, it's probably pretty helpful to come from a specific niche.
- [79:30]
B
Totally. So my break into AI policy, I mean, I'm what passes for an old timer in AI policy because I've been doing this for 10 years, beginning with my measly internship at the White House Office of Science and Technology policy back in 2015. And so AI policy was already a thing in 2015. The White House wrote three reports on it. So when I was trying to get people to take me seriously, I was like, well, they've got Ed Felton, who is the Deputy Chief Technology Officer of America, and he's also, you know, a very distinguished professor, professor at the Princeton, you know, computer science department. The stuff he's writing about, probably nobody cares what I think, you know, on those topics. But what I had found out was that Ed Felton, you know, was charged with writing three reports. The Obama administration was originally going to write a fourth report titled Artificial Intelligence and National Security. And then they decided that they would punt and that the, you know, the Hillary Clinton administration would do that back when they thought Hillary Clinton was going to be president. Right. So there was this report that was not written on artificial intelligence and its implications for national security. And I was like, oh, maybe that could be my master's thesis. Right. Like the report that was not written. Because here's like a niche where there's obviously demand but there's no supply. I mean, like at the time there was decent work that was being done on autonomous weapons at this point, but sort of like a broad based survey of progress in machine learning and what it meant for AI was a pretty unplowed field at the time. And so I was like, okay, I'm going to find a niche and try and write the best paper I possibly can on this niche where there is demand but there is no supply.
- [81:31]
A
Yeah. And it is, it is wild that AI national security was a niche just like 10 years ago.
- [81:38]
B
I know, but here's the thing. Because, because it was a niche, somebody like me, who had never written anything as a think tank scholar in the past could get permission to write it.
- [81:50]
A
Yeah.
- [81:51]
B
So like what? I mean, like, like the perfect career. Like, I mean, I've said before, like, I've gotten lucky many times in my career. One of the ways in which I've gotten lucky is I broke into the field by competing in a niche where the competition was relatively weak and the demand was high and growing. You know, so like now artificial intelligence, national security is like a huge topic, right? There are entire think tanks where all they do is AI and national security. And now that the competition is, you know, much tougher. Well, I'm also more competitive, right. Like, if I had to break into AI and national security today with, you know, what I was bringing to the table 10 years ago, it'd be a lot harder. And so I would probably be asking myself, like, what's, you know, what I wrote the general survey of AI and its impact on national security. That's like a terrible paper to be somebody's first paper right now, like, if I was writing my first paper, I'd be like, what is a niche of, the niche of artificial intelligence national security, where I can make a genuine intellectual contribution. Right. You want to find out like, where, where can I do something that is new and interesting and people will find useful? And you know, there's. When I was in business school, they taught about the sort of typical career path which they called like an. This is a typical very successful career path which they called like the hourglass, meaning when you start your career at the top, you're very broad. And then, you know, because you have like what, like say like an undergraduate degree in economics, and then you specialize a little bit more, a little bit more, a little bit more, a little bit more until you do your PhD and then you're like, like super hyper specialized in whatever the thing is, your PhD is. But then you Know, once you have a doctorate in economics and are a fancy economist. Right. That's. And then permission to start branching out again because people, you know, will find interest in that. And so I do think that, like, you know, pursuing specialization maybe not as like a permanent phenomenon, but like, as a way to identify where can I make a contribution? Is usually a pretty good thought.
- [84:22]
A
Yeah. What other skills and subject matter expertise do you think are the most important to have when you're trying to pursue a career in AI policy?
- [84:31]
B
So, you know, I'm not an engineer, I am not a good programmer. Nobody's going to hire me as their computer programmer, although I can, you know, program.
- [84:44]
A
They don't need you anymore. They have AI.
- [84:46]
B
Yeah. Bingo. Oh, yeah. I'm definitely not as good as AI. Like, yeah, yeah, like I said, nobody's gonna hire me. But I do think that, like, there's no substitute for being willing to learn the technical complexity of the topic. Right. I mean, I, there, there are aspects of learning in AI that I strongly suspect are pretty difficult to understand without, like, being a part of the team building it or operating it. You know, I think there's probably some like, layer of understanding that is forbidden for somebody like me, but there's many, many other layers of understanding of technical nuance that are, you know, addressable to someone like me if you're willing to put in the time and effort to actually learn how this stuff works. And so, you know, I wrote a paper that came out in April 2019, you know, called understanding AI technology. And it was, you know, just describing, like, what is supervised machine learning, what is unsupervised machine learning? What is reinforcement learning? You know, if you wanted to have a product that was using these things, like, what are the different aspects that you would need as success criteria to be able to run, you know, a supervised learning algorithm and build a AI model that's actually going to be useful to someone. And like, like, I actually, I actually, But I think I said it in the paper. Like this paper, you know, constitutes the minimum degree of technical nuance that you need to understand to not make terrible decisions as like an AI policymaker.
- [86:23]
A
Yeah.
- [86:24]
B
And I, I think that's like always worth sort of like asking yourself as you're, as you're involved in science and technology policy, which is like, okay, I'm, I'm, I wanna, I wanna make a contribution here. It's in a science and technology field, obviously. Like, I don't need to, you know, be the chief engineer of the company to have standing, but if I'm oblivious to key technical and political realities, then I'm going to make dumb mistakes. And so you just sort of have to ask yourself, like, what is. How do I learn the minimum degree of technical nuance? I need to. To make sure that, like, my arguments are grounded in reality and plausibility and coherence. And I think it kind of just comes back to, like, reading and specifically, like, reading with the. There's a. There's a physicist who won the Nobel Prize called Richard Feynman, kind of a hilarious character. His memoirs are quite funny. But he, you know, had this thing which other people. I don't think he called it this, but other people have called it the. The Feynman learning method. And it's about, like, how you don't really understand something until you can explain it to other people and then they understand it, right? And, like, when you. When you think you understand something, you should constantly be looking for edge cases because the edge cases will highlight to you the ways in which you do not understand the things that you think you understand. So, like, this is not a perfect example, but, like, what is the relationship between water and fire? Right? Like, 99% of people right now are thinking, well, water puts out fire, right? But what are the edge cases? What if it is a grease fire? Well, if you pour water on a grease fire, the fire gets bigger and it spreads and it's terrible. If you pour water on, you know, an electrical fire or electrical socket, you may start the fire, not, you know, stop the fire. And so you, like, think about these edge cases, and it deepens and expands your understanding of the topic. And so, like, when I am thinking about earlier in my career, rocketry and rocket engines, which I had to spend a lot of time doing it, I'm never going to be a rocket engineer, but I had to write corporate strategy memos that were grounded in the reality of rocket engine manufacturing and the industrial economics associated with that. And so what is all the minimum degree of technical nuance that I needed to understand to know, not be erroneous? And like, when I thought I understood something, I was constantly, like, asking myself about edge cases of, like, well, would it be true if? Would it be true if? And again, you know, as you constantly hunt for imperfections, that's where you get closer to perfection. And like, I've never achieved perfection in my knowledge or understanding of anything, but that is, like, how you get closer. And so, so I'm constantly looking for gaps in my own understanding because that tells me, like, oh, what the next Thing I should read is sure I.
- [90:03]
A
Want to next talk a little bit about breaking into the field of AI policy. A fair amount of our listeners probably already work in some part of the AI policy field, but a substantive amount are currently not working on AI policy and might want to be working in AI policy sometime in the near future. I think a lot has changed about getting into AI policy since you got into AI policy. But I'd like to hear just like your top line advice for if you're someone say like mid or early career working in consulting or engineering or something else, what is the best way to break into AI policy?
- [90:41]
B
Yeah, I mean as you said, it's different than when I was doing it, but I think it, it the, the starting, the starting point is the same, which is asking yourself, how would I be useful at AI policy? Right? Like how would I do I currently possess and if not, how could I acquire the skills that would make me useful as somebody working in AI policy? And so like I often, that's like the high level advice. The second thing that I have done at many, many times in my career is like LinkedIn stalking.
- [91:22]
A
That's how I found you.
- [91:24]
B
Yeah, like this, this is what I did, you know, before I was applying to graduate school. And it's definitely my advice for anybody who is considering graduate school, which is do not go to any graduate school where you cannot identify people who graduated from that school who have jobs that you want. But like, and I learned that by like LinkedIn stalking. I was like, okay, here's people in the world who have jobs that I think I might want someday, like, what was their career path like? Like how did that work? How did they break into that field and sort of working backwards from what skills did they have, what experiences did they acquire? And then thinking about, you know, okay, so how do I become that kind of a person person. So that's thing number one is like deserve to work in AI policy. The second thing is to like have proof that you would be useful. So I mentioned, right, like how I had my master's thesis, this published report, AI National Security. And I used that to get myself a non resident fellowship at a think tank. I was like, hey, here's the kind of work I'm capable of doing. You know, if you hire me in this case it's like unpaid because it was a non resident fellowship. But if you hire me, I'll do more of this. And that's like the second thing. Then the third thing is like about all the different kinds of institutions where you can do this work. And the reality is like, AI policy is a pretty diverse community now. It includes NGOs, it includes Congress, the executive branch, executive agencies, international organizations, companies, on and on and on. And I'm probably not going to do a great job of like detailing all of those different situations and for whom they might make sense at what stage. But Matt, you have told me about a resource that was actually created by a CSIS alum, Remco Zwetsloot now at the Horizon Institute, I think it is.
- [93:26]
A
That's right, yeah, yeah.
- [93:28]
B
And that is emergingtechpolicy.org and so they have a bunch of written out career guides, which is just like how to work at a think tank, how to work in Congress, how to work for an executive branch agency, et cetera, et cetera. And they have that for AI policy, they have that for space policy, a bunch of other stuff. And I haven't read it all, but the stuff I read looked pretty reasonable.
- [93:52]
A
I've read quite a lot of it, especially for the AI policy side of things. And so I will vouch for a lot of the stuff on there. I think there are also some really useful databases of fellowships. I think, especially if you're a technical person trying to get into AI policy, fellowships are a great way to do that. And emergingtechpolicy.org has a whole list of all the fellowships where they're especially looking for technical people trying to work at the executive branch or in Congress and whatnot. So that's another part of it I'll highlight.
- [94:22]
B
Yeah, sounds great. What else?
- [94:26]
A
Yeah, I think as someone who just supposedly broke into AI policy, one might say I have maybe one top line thing for when you're applying to jobs that I feel like is really important to share and sort of goes back to the story of how and why I reached out to Greg. So again, in defense of LinkedIn stalking, I think if you're applying to opportunities and you're just cold applying, it turns out that there are a lot more people that want to work in AI policy than there are jobs in AI policy. I know this is true for a lot of parts of policy in general, but AI is a particularly hot topic. And so if you're cold applying to jobs, you're probably wasting your time. I was probably wasting my time. I spent a lot of time cold applying to jobs feels very productive. Unfortunately, it's not. I think there are like, there's a pretty clear hierarchy of ways to find job leads. And by that I mean anything that can turn into A job. And most of the time, anything high up on that hierarchy is going to involve knowing someone at the place you're applying. And it can be hard. It can be hard to find someone. The first email I sent to Greg did not get a response. But the second time, as you said, he connected me with someone else in the program. And that conversation probably played. Played a big role ultimately, in helping me get this job. And so I think it's important here, going back to the 8020 rule, which also applies to being efficient with your work. Maybe the 20% of your work that you're doing is responsible for 80% of the outcomes. You want to focus on the most efficient 20% of ways that you can be applying to jobs here. And that's not going to be cold applying, as appealing as that might have been, be as. That might be as, like, productive as it might seem. By far, the best way to get into an organization is to talk to people in that organization in the specific program that you want to work in, learn from them, and also impress upon them that you would be good at any role that they have open and that they're hiring for.
- [96:30]
B
No, I don't have anything to add. There you go. Sound advice. Thanks.
- [96:34]
A
Okay, let's move on to a few other topics in the AI policy career advice ecosystem that I want to touch on. On one that is a little specific to AI policy is how to make decisions about where to go given the pace that the technology is advancing. I think this is something that a lot of people, including myself, think about. Depending on how good you think AI will be in one year, in five years, in 10 years, in 100 years, the decisions you make to have an impact in policy would look pretty different. And unfortunately, there is no consensus on how good AI will be in any of those timeframes. And in fact, it's quite the opposite. You have people who say AI is going to be spectacularly better than humans at everything in two years, and you have people that say that will never happen. How do you suggest people price these predictions into their career decisions?
- [97:29]
B
Well, we're kind of getting back to the AI and Labor podcast episode that we had with Barry Holzer. The former.
- [97:36]
A
That was a prime example of different predictions.
- [97:39]
B
Yeah, yeah, yeah. That he was a former chief economist at the Department of Labor. And so I think it. It in principle. Right. Makes a lot of sense if you are, you know, Dario Amadai and you're saying, I. I forget what his exact quote was, but it was something like, you know, 10, 10 to 50%. Somewhere in that range of like, entry level white collar jobs are going to be automated within the next, you know, handful of years. Let's say that's true. Yeah. That should definitely, you know, have implications for how you think about your, your career. The my sense and, you know, my opinion is one opinion you certainly don't have to believe it is first that it's going to take longer to get to quote, unquote, AGI than that prediction. You know, Andre Karpathy, who's the former chief AI officer at Tesla, who is also, you know, one of the original staff at OpenAI and is a lovely professor of AI in the past, you know, he just did a great interview with Dwarkesh Patel and he was talking about how he thinks, you know, just based on his instincts, having been in this field a long time, seeing the pace at which progress advances, seeing the challenges that currently, you know, stand up, you know, thinks he's like, we're, we're a decade or more away from that. I find that credible based on just my own experience, my own understanding of the field, with the caveat that I understand that I could be wrong. And that's why on this podcast we always talk about the diversity of prediction. But I want to talk about something else, which is chess.
- [99:33]
A
Chess.
- [99:34]
B
So in the case of chess, computers have been better at chess since 1997. What some people might not know is that the best chess player on the planet between 1997 and I think it was 2015, maybe 16, maybe 17. I forget when it was. But the best player, even though the best computer was better than the best human, the best chess player on Earth was not a computer. Between 1997 and 2015, the best chess player on Earth was a team of a chess grandmaster and a chess playing computer program. The human working with the computer was better than the human alone, and it was better than the computer alone.
- [100:27]
A
And they call this cyborging, I think.
- [100:28]
B
No, they call it. The term of art was Centaur chess.
- [100:32]
A
Oh, Centaur.
- [100:33]
B
And the tournaments were called, like, freestyle chess, but centaur because, you know, it's like half horse, half man.
- [100:38]
A
Yeah.
- [100:40]
B
And so my point here is that my hypothesis anyway, is that even as we continue to make AI progress, even as we continue to make incredibly impressive AI progress, there is going to be this period of time where top performance continues to be humans working alongside AI. Now, like in a thousand years, do I expect that humans will be contributing to the economy in the same way they do today? No, certainly not. Right. Like eventually, I expect AI to surpass human intelligence in every way. And that's just because, you know, the human brain is 2-3 kg of gray matter running 20 watts of electricity. That's, like, by far the most impressive computational device that has ever existed. But there's no law of physics that says that it has to always and forever be the most impressive computational device that has ever existed. And so at some point, you know, I do expect the march of technological progress to include superhuman AI. But the point is, as in the Chess example, even after we have superhuman AI, it doesn't necessarily we have the, you know, that humans are irrelevant. And then if you ask yourself which humans are going to be relevant, my hypothesis is it's going to be highly skilled humans who are still relevant. And so that just comes back to, like, all the advice that I gave you before, which is like, are you acquiring awesome skills? Are you thinking about, like, how you can be relevant in the marketplace? And so my advice to the people who have a short timeline, and my advice to the people who have a long timeline is the same, which is, go become a very valuable person. Like, you know, go be the type of person who should obviously be hired for your dream job. You know, become that person. And I think that's true, regardless. Now, the other thing I'll say is just like, we've talked a lot about, like, practice and improving, and I'm certainly in the camp that thinks that AI is very useful as a research supplement. You know, there's certain kinds of information where a deep research query on either Google or on OpenAI gets me better results than a Google, a standard Google search query. But, like, if ChatGPT is doing your writing for you, you're doing it wrong. And you will not learn how to write or even how to think by outsourcing that function to chatgpt. So the point is like, like, these are definitely useful tools that you should learn how to use, but you also have to learn how to do stuff yourself without them.
- [103:49]
A
Yeah. Well, one more topic I want to talk about before we wrap up is grad school. You hinted earlier in the conversation that we were going to be talking about this, particularly in the context of AI policy. I'm curious how important you think grad school is and whether it's something you would recommend.
- [104:06]
B
So I think grad school is potentially good. I don't like, I went to grad school. Not everybody who I went to grad school with was making a great decision by going to grad school. Costs a lot of money, takes a lot of Time. I mean, that's time you're not making money and it's time you're not doing on the job. Learning, learning. And as I think about my educational experiences, like, my most formative educational experiences were debate, the magazine, working at Avocent, and within grad school, my most formative educational experience was actually the master's thesis. Not like the classes, although I had some classes that were really wonderful and changed my life. Like I mentioned Jeff Seglin and the writing class.
- [105:00]
A
Sure.
- [105:02]
B
But, you know, you should. You should go to grad school with, like, a theory of what you're going to get out of it. And here's what I would sort of tell. Here's what I would say are, like, the criteria for how to know if. If grad school even might be a good option for you. Number one, do people who are getting jobs that you think you might want go to that graduate school? So I'd be like, you know, what is a job that I think I might want to have five years from now? Go find the people who have that job. Did they go to grad school five years ago? If the answer is 100% of the time, no, probably don't go to grad school.
- [105:38]
A
Yeah.
- [105:42]
B
The second question is, I would never go to graduate school without knowing already, before I even apply, if I went to this graduate school, who are the professors that I'm going to want to form a relationship with? Right? Like, when I applied to the Kennedy School, I was like, I can't wait to work with Nick Burns, the former ambassador to NATO, the former ambassador to China. I can't work, wait to work with Joseph Nye, the former head of the White House National Intelligence Council. And like, for me, getting into the school was just like the first application gauntlet. The second application gauntlet was like getting these professors to care about me and take me seriously as one of their students. Because, like I said, you know, a professor. A professor, depending on the class, might have 10 students, they might have 60 students, they might have 300 students, right? I promise you, they are not opening their Rolodex and making generous introductions for all 60 students or for all 300 students. So it's really about, like, making a good enough impression and investing in the relationship and going to office hours so that, like, you know, when I was writing my master's thesis, Nick Burns, he. He wrote intro emails for me to meet with the former head of, literally, MI6, like, the British intelligence services. MI6. I got to meet an interview. I got to interview the head of emerging technology policy in NATO. Like, those are introductions that Nick Burns made for me. They meaningfully affected, like, the course of my career in a wonderfully positive way. And the point is, like, that's some of the biggest stuff I got out of graduate school. And classes can be awesome, can be lame, but, like, I just. I would not recommend anybody drop hundreds of thousands of dollars in years of time going to graduate school just because it seems like a thing a lot of people are doing. It can be a great choice. It was a great choice for me, but I had a plan.
- [107:59]
A
Yeah. Yeah. Well, I have one closing question I want to ask you, but before we get there, is there anything else that you want to share about advice for AI policy to.
- [108:09]
B
No, no.
- [108:10]
A
Fair enough. I mean, we're closing in on two hours, so. Talked about quite a lot if anybody.
- [108:14]
B
Made it this far. Yeah, thanks.
- [108:16]
A
All right, well, my last question here. If you could go back and talk to your. Your high school debate self. High school debate, Greg, what is the one piece of career advice you would give?
- [108:28]
B
I mean, I think, like, all the advice that I've. I've given throughout this are things that I, like, learned over the course of my career. So if I could go back to high school, Greg, I would probably just cheat and try to give him all of this advice without having to learn the hard lessons over the course of the career. But I think if I was to advise a high schooler or anyone else today, the habit that fortunately debate sort of already brought out in me is one that served me very well. And if I was to, you know, advise a high schooler today, I would just do this, which is just admire great work, but don't stop at admiring it. Try and deconstruct it. Try and figure out what makes it great. And so for me, you know, in debate, that was like being a freshman, looking up to the seniors like Femi Moranfala, who was amazing and being like, why is Femi Moranfala amazing at debate? And, you know, what could I do that would make me more like him and less like the bad performance.
- [109:35]
A
Yeah, the fact that you still remember his name is clearly important.
- [109:38]
B
He was my hero. He was a senior when I was a freshman, and he was just a brilliant speaker and thinker and had a hilarious hairstyle, which was funny. So, yeah, I think admiring great work and deconstructing great work and then finding ways to put in, like, formal practice, I think. Is it so, like, one thing that I read a biography of Benjamin Franklin a while back and something I learned in that biography that kind of blew my mind is so Benjamin Frankler was like a newspaper editor in colonial America and is now regarded by historians as probably the most talented American writer of his era. By the time he was like, 22, like, he's the best writer in America as a 22 year old. And, you know, the question is like, well, how did he become the best writer in America? It turns out that he had a daily writing practice habit. And specifically what he would do is he would take out, you know, an essay that he admired, and he would convert the prose of the essay to poetry, and then he would hide the original version of the essay and then use the poetry translation to create a new prose essay. And then he would compare his prose essay with the original prose essay and be like, okay, what? What did they do? That's better. What did I do? That's better? Et cetera, et cetera. And he did this every day, you know, and the point is not that this is the world's greatest writing drill that everyone should do. I don't do this writing drill. But I just don't think it's an accident that the guy who was the best writer in America by age 22 is the guy who had a crazy daily writing practice routine. You know, practice is awesome. And if you can carve out time in your schedule to do it, you can be awesome too.
- [111:37]
A
Yeah. Well, Greg, I have found this immensely helpful. I really hope our audience has found it equally helpful. Thank you again for taking the time to share all these ideas with us today.
- [111:48]
B
Thank you, Matt. I really enjoyed the conversation. Thanks for working here.
- [111:52]
A
Cool.
- [111:56]
B
Thanks for listening to this episode of the AI Policy Podcast. If you like what you heard, there's an easy way for you to help us. Please give us a five star review on your favorite podcast platform and subscribe and tell your friends. It really helps when you spread the word. This podcast was produced by Sarah Baker, Sadie McCullough and Matt Mann. See you next time.