I gave AI a year of my life: tech journalist Joanna Stern

0:00Hi, everyone. It's Rana. The future of AI is not built in isolation. It is shaped in rooms where ambitious people challenge each other. Masters of Scale Summit is back October 20th through 22nd in San Francisco, bringing together founders and innovators pushing the edge of what is possible. If you're building what comes next, you should be in this room. Apply now at mastersofscale.com slash pioneers. That's mastersofscale.com slash pioneers.

0:50Airbnb, Zillow, Microsoft, Liquid Death, and more. Hear from the founders who've changed the game. It's anything but business as usual. Find Masters of Scale on Apple Podcasts, Spotify, YouTube, or wherever else you get podcasts.

1:10Every Sunday, we cover the week's tech news on This Week in Tech. Hi, this is Leo Laporte inviting you to join me this week as Berber Jin from The Wall Street Journal and Paris Martineau from Consumer Reports join Ian Thompson. And we'll talk about, of course, OpenAI and Anthropic. They got together with a bunch of religious leaders and decided what religion AI is. They've also figured out how to keep it from blackmailing you. You just say, well, that would be wrong. This Week in Tech, you'll find it at twit.tv and wherever you get your podcasts. Hi, everyone. It's Rana here. Before we hop into today's episode, I'm joined by Sharif Mansour, head of AI at Atlassian.

1:51Atlassian creates productivity and collaboration tools like Jira with AI-powered Rovo to bring humans and AI agents together. They also support and sponsor Pioneers of AI, something we deeply appreciate. So Sharif, I am so curious. When the newest member of your team isn't human, what changes? I think a whole lot of dynamics change on how we collaborate together. Teams often have to design their personality, how they want the agent to interact. Is it a customer service, an internal agent? Is it a bank? Is it a media firm that's doing creative marketing? That's all in the voice and tone of the agent.

2:29But a lot of teams out there in businesses are putting agents in an actual business workflow. So when something happens, the agent picks up the ticket, does something, passes it to the human to review it or that kind of thing. What are the top human skills that we need to learn to be awesome teammates for AI agents to work productively alongside AI? Yeah, two immediate ones that come to mind is the skill of just being curious. And the second one's really being able to articulate really well, such as I was talking to a customer the other day that was using an agent in a procurement workflow. So these are teams in a business that get contracts and they have to review the contracts And the agent does an amazing job of doing a first pass, kind of like finding a needle in a haystack and saying, I think Section 3 is going to be a concern for you.

3:17And Section 10 didn't meet this criteria. In this case, they're bottling their workflow in JIRA, they pass the JIRA ticket to the human, and the human will review it and go back and forth. So the better you are at articulating the specific role and how you want the agent to work, the better the outcome. Do you have any best practices for how companies should think about human oversight? Right. Some criteria that might help you do that is understanding what are the safety concerns that might occur from this business workflow? That might be a good signal of where you might want humans in the loop.

3:47The second area they might want to do this is compliance. A lot of our customers have highly regulated environments, whether it be banking, finance, telcos. Value is probably the biggest one. Where is the highest value that you're getting from your team? And how do you enable them to do more of that? So is my value in getting a help desk ticket and that ticket is another password reset and I'll manually do the password reset myself and I get like 5 ,000 of them a day. And so we would go, okay, how can we hand that off to an agent in a safe way, but also in a way that makes sense for our customers that still meets their needs.

4:23What's the one takeaway you want to leave us with today? I would say the world loves celebrating the individual over the team. We all know that's actually not true to an extent because the team is what elevates the individual. Look at the example of LeBron James, incredible basketball player. He's only as good as the team around him. A lot of the demos we see are about individuals making amazing work with AI. And that's awesome. But I would encourage everyone out there to think about how is AI and agents elevating the whole team? Because when that happens, the outcomes of getting a diverse team, working together with humans and agents climbing together is far better than celebrating an individual, one-shotting an amazing prompt.

5:05My encouragement to everyone is just think about the team of an individual. And I think that same logic applies for applying AI to our teams rather than applying AI to our individuals.

5:17The students graduating this month in May or June, when they were freshmen, ChatGPT came out. And they have seen their reliance on this. You know, the models get better. They've gotten more reliant. You alluded to one particular section where you talk about your Waymo adventure. So tell us about that. The kids and my wife, for the whole week, we only ride in Waymos. But I got to see this through their eyes. Like, they did not care at all that there wasn't a human behind the steering wheel. Versus my wife, who was basically crying in the back when, you know, we first get into the car. Most of the time I say please and thank you.

5:57I will sometimes get quite mad at, like, these AIs. Where did you land on that? Like, when Claude has gone off and built this whole website or built this whole agent that saves me two hours at night from having to manually do the work, I genuinely am so happy. And so I praise it. Like, I'm like, that was amazing work. Great work, right? It is not easy to cover AI as a journalist. There's a constant drumbeat of headlines. What approach is to learn by doing? to try all the new technologies yourself. That's the specialty of Joanna Stern. The longtime tech journalist just left the Wall Street Journal to start her own media brand called New Things.

6:39She also has a new book, I Am Not a Robot, which I read and absolutely loved. I highly, highly recommend it. It's all the stories and lessons learned from Joanna's year of trying pretty much every AI consumer product out there. And her insights are as sharp as they are funny. I'm Rana El-Khalyubi, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

7:11Hi, Joanna. Welcome to Pioneers of AI. I'm so excited to have you on the show. Thank you so much for having me on the show. This is exciting. I guess first-time guest, long-time listener. Yay. It just took writing a book, a small little book, to make this introduction happen. Yes, just a small book, which I have right here. We're going to talk about it. But I want to start by your big pivot. So you have been a tech journalist at The Wall Street Journal for, what is it, over a decade? Yeah, it was 12 years. And you just left and launched your own consumer media company called New Things. Yep. So tell us about that.

7:52Like, what was the impetus for making that big shift, especially now where, you know, the media landscape is kind of being disrupted and it's not easy to find these big roles or big jobs? I'd been at the Journal for 12 years. I had done so many amazing things there and had so many great editors and peers and just great teams to work around. And I did things like incredible videos, hosted events, wrote the weekly column, and I was ready for something different. There's so much happening in media right now. But then to bring it back full circle to the book, the prevalence of AI tools and the ability to build a business now and be more efficient at it, absolutely played into this.

8:38There's so much underlying infrastructure and things that you need to get done when you're building a business. And I saw this when writing the book that I was basically creating a small business around this book, right? I was hiring people. I had expenses related to it. And so you sort of learn to manage this, but you also learn that you can be so much more efficient. The first page of the book is actually how I use AI to write this book. And AI did not write the book. But there were all these places where it intersected with research, with being able to put together spreadsheets, to be able to get things done more quickly, where I say in the book, I actually was able to write this book, I think, two years faster than I would have been able to before, right?

9:21So that experience, in combination with where the media industry was going, in combination with the fact that I was just ready for a change after 12 years, here I am building something new. And we just launched, it's called The New Things. You can go to thenewthings.com and we've got newsletters and videos and hopefully events in the future. And the goal is really just to help people, guide people through interesting tech stories with a journalistic lens, with also humor, because I'd like to do that in my videos. And hopefully it works out. Okay. So I do want to talk about the book, I Am Not a Robot.

9:57First off, Congratulations. Thank you. I could not put this book down, but my most common comment in the book was OMG and LOL. Lots of LOLs. It's hilarious. So I like really, yeah, it's really awesome. But also. Thank you for reading it. Why write this book now? You've been covering tech for years. So what's special about this AI moment that kind of provided the impetus for writing the book? Yeah, I wanted this book to be for the people who know that AI is important to know about or are already using AI, but want to know about this future where we keep hearing about from every executive that AI is going to change every part of our lives.

10:37And so I sort of challenged myself and said, okay, you guys are all saying this. Let me go try this. Can I do this for a year? In the sense that like this AI moment where people say it's going to change all our parts of lives, shopping, medical, transportation, education, socialization, relationships. Can I try as many of these things and see if it's really going to change our lives? And so that was what I tried to do, live in the future, which is really hard because we obviously live in the present. But I just tried to try as many things as possible and see if I could get there. Okay. So you are very transparent about how you use AI to actually write the book.

11:18So all the words are your own, but you did enlist AI to do interviews and kind of synthesize research and whatnot. I think one of the biggest fears people have about AI is around job displacement. And you definitely explore that theme throughout the book, not just in how you wrote the book, but also in kind of thinking about AI doctors and therapists and even like romantic partners. Um, what is your TLDR on kind of this whole AI and how it's displacing or not displacing jobs? Yeah, I mean, look, there's so many parts of this book where like, look, you have a weekly show about AI and you can barely keep up with everything, right?

11:59But I think I got a lot right in this book, especially about this next generation and their struggle to get jobs. There is this chapter where I actually go work at a customer service center and okay, I really see what's happening there. Those jobs are going to AI and they need less human customer service agents, right? But then you see it happen differently in, say, the radiology or the mammogram chapter where the point wasn't even about the jobs. The point was really about, are we going to trust AI to make medical decisions for us? And I go and I have my mammogram and ultrasound read, but I sit side by side with a doctor who says, I love AI.

12:35I love that I can use these tools, Transpara and ScreenPoint is what they're called, and they give me guidance as I'm working with it. And so, you know, Jeffrey Hinton famously said, like, why would you even be a radiologist? They're clearly going to be out of work. And that's not where we're at right now. We see these professions being enhanced by these tools. So it totally threads through the whole book. I think one of my favorite, you mentioned it, is the massage chapter. Have you had one, by the way? I have not. Should I? You got to do it. You got to do it. Okay. Okay. Okay. I'm going to try it.

13:12It's just worth trying because it is so different than being with a human massage therapist. So like I kind of sitting there on this massage table thinking to myself, well, what's this going to do to massage therapists? And then you start digging into it and you're like, okay, there's actually a shortage of massage therapists. And some people don't even get massages because they don't like to be touched by humans and this could solve that problem. So everything is so nuanced in different professions, and that, I think, is a big takeaway of the book. You know, I've become fascinated with this idea of when you are delegating work, how do you decide whether to delegate it to a human or to an AI?

13:50Do you have a point of view on that? Oh, I love that question, and I should read or listen to more of you wondering on that because I have been starting this company, and I have a very small team. They're a team of three. It's me, my video producer, David Hall, and then we have a production assistant whose name is Amaya Austin. And this is an amazing team of humans. All humans. Right. But everyone has AI, right? When I hired Amaya, in fact, she's right out of school. And I said, Amaya, every day you need to be interacting with AI. We are a startup. We have a ton to do. And there are things I'm going to ask you to do.

14:29You go do it with AI. That's fine, right? Obviously, I don't want writing. I don't want editing being done by AI. But in terms of like editorial, to some degree, like we have some nuances there. But I'll say to myself, OK, do I ask Amaya to do that or do I just like go off and do it right now? You know, have Claude Code go and do it for the next 15 minutes, right? But I also know that Amaya might go make that same decision, right? She might decide I might personally do this or I'm going to have Claude Code do it. And I can look, look, the truth is I can tell, but I also don't really care. I wonder what you think of this.

15:08Do you have a team that you sometimes have to delegate things to? Yes, we are also a small team in my fund. And we have a couple of like junior analysts. And I found myself like more often now asking myself the question, do I delegate this to, you know, to these folks or should I just have Cloud Code do it? And it's very intuitive right now. It's not like I have a checklist to decide how to delegate, but I'm noticing it more often. It's interesting. Well, I do feel this opposite feeling, too. I don't think people talk about it enough, and I don't know if there's a name for it, but when AI does something really good for you and it does it, like, you're like, do you ever feel proud of it?

15:51Yes, totally. I'm like, yes, right? I don't know if there's a name for this. Our chief of staff AI agent, we call her Blue because the fund's called Blue Tulip Ventures. So she's BLU. And I'm like, you go blue. I'm like, amazing. Right. Yeah. Right. It's different. It's that different satisfaction of like, yeah, you go AI. That doesn't feel anything probably when I tell you great work. Like, you know, made the agent to collect the email addresses and put them into a spreadsheet. And so I don't have to. Like, you're just like, oh, that's so wonderful. But the feeling of when a human does something and does it so well and like you're just so proud of the team in general is a different feeling.

16:33It's a different kind of pride. Yeah. But I think that you feel even more of that now given that these higher level tasks can be done by the human. You know, I think there's just this dichotomy now of the way these tasks get done. Coming up, we'll talk about AI and good manners. That's after this short break.

17:12Whether you rent or own, your housing payment is probably your biggest bill of the month. So here's something worth noting. BILT is the membership that actually rewards you for it. Every housing payment earns you points that you can redeem toward travel with partners like United, along with stays at Hyatt Hotels, Lyft Rides, Amazon.com purchases, and more. And the Neighborhood Concierge feature can make restaurant reservations, book fitness classes, and find you local spots, all while racking up rewards at 45 ,000-plus merchant partners. So whether you're paying rent or paying a mortgage, BILT works for both.

17:47Join the membership for where you live at joinbuilt.com slash pioneers of AI. That's J-O-I-N-B-I-L-T dot com slash pioneers of AI. If you've spent any time building AI products or leading technical teams, you know this. Transformation doesn't fail because of ideas. It fails because teams can't move together. Enter Atlassian's teamwork collection. It has planning in JIRA, documentation in Confluence, video updates in Loom, and now AI agents in Rovo, which connects the dots across your work so nothing gets lost. It's one AI-powered teamwork platform designed for how modern teams actually build. Learn more at Atlassian.com slash TeamChanger.

18:39That's A-T-L-A-S-S-I-A-N dot com slash TeamChanger.

18:50On April 4th, 2023, around two in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundry, the killing of Bob Lee. beginning April 16.

19:20You know what also struck me in the book, kind of your pondering around the etiquette. Oh, yeah. The etiquette around kind of dealing with AI. Like most of the time I say, please and thank you. I will sometimes get quite mad at like these AIs. Where did you land on that? I'm nice to AI. I mean, yes, you're talking about this section in the book where I went and go to talk to the Emily Post Institute. I went to go talk to Daniel Senning Post, who's the, I believe, the great, great grandchild of Emily Post, you know, the famous etiquette guru. And he was so honest. He was like, look, it doesn't matter.

19:55Sure. Like, but he did say he was he was very educated about AI. And he does say it does matter in some places where the learning when you say thank you, when you praise it, well, then it knows it's done good. And so it might do more of that work. Right. And it's good for the training. So we obviously know that. But he's like, look, does it matter? No, but it's more about you and your feelings and your affect after doing this, right? Like when Claude has gone off and built this whole website or built this whole agent that saves me two hours at night from having to manually do the work, I genuinely am so happy.

20:28And so I praise it. Like, I'm like, that was amazing work. Great work, right? I don't know. I think it makes me maybe a better manager to humans too. Because like I said, you know, Amaya's going to listen to this and be like, oh boy. You know, and you also feel good when you hire a human who's just great, right? I mean, that's part of managing. And so I feel great about hiring Amaya. She's amazing. If someone's listening to this, if you try to hire her, I will, yeah, I'm going to knife you, okay, with my book. I'm going to come after you and hit you with books, hardcover books. But there's something special about that, too, like knowing that you have put someone in the right role, that they are thriving, that you can see so much potential in them.

21:06And that's my biggest worry, you know, every day about, especially in journalism, this younger generation, because so many of the tasks that Amaya once would have been doing, I do have AI doing. So every day, we have so much for her to do, but I am very strategic about what I have her do. And a lot of people in the workforce are thinking that way now. And that's the fear for this new generation, you know, as I call it in the book. Gen-gen, right? Generation AI. Like, this is the students graduating this month in May or June are the first college class to have gotten ChatGPT. When they were freshmen, ChatGPT came out.

21:45And so they have seen this progression and they have seen their reliance on this. You know, the models get better. They've gotten more reliant. And so we're going to see, I think, a lot happen this year. You know, it also made me—I loved reading the book. but there is a version where I would upload, you know, just ask ChatGPT for a summary of the book, right? Yeah. And I was kind of pondering about that. And I think we lose a lot when we do that because I feel like I just know you now so much better. Like, for example, you talked about The Lampoons and Honey, I Shrunk the Kids. And these are some of my favorite movies growing up.

22:22And so when I read it, I was like, yes, go Joanna. I just like, I felt this personal connection to you. I'm pretty sure an AI summary would have skipped that line because it's not like that core to like the book, right? No. So what do you think of that? It's funny because I hired a human editor, funny enough, coming. I mean, and he was amazing because I realized I really need someone to, you know, brain me in. And the first draft of the chapter where I talk about National Lampoon's Vacation, which is the chapter about Waymo, I probably went on for like three pages about that movie. And he's like, that movie has nothing to do with this chapter.

22:58And I was like, I am Chevy Chase. I am, you know, Clark Griswold and I'm bringing my family. And he's like, yeah, in and out, you know, like get in the car, get in the Waymo and speed this along. I totally agree. And I've really grappled with this. If I've known you've bought the book, I would really love to give you the option to talk to AI about this book. Oh, your AI, your digital twin? My digital twin or a version of a bot that I'm able to train. And then there's the fear and, you know, probably some fear on the publisher's end of, okay, well, first of all, how do we know that these people bought the book?

23:35And second of all, is this the way you want people to interact with the book? My feeling is like, look, I have a ton of nonfiction books. I buy them all the time. I buy them for my peers. I buy them because they look great. I buy them because I read the review. I don't often read the whole book. Really, I have to say. Like, I'm admitting it here on this podcast. I often read the first third, then I'll go back. And I've designed the book to be that way in this with the short, like, I have these journal entries. I have these little snippets, these Q &As. And I wanted it to be that. So my hope is that one day there is a better way to have AI interact with books, but that it doesn't take away exactly what you're talking about, which is like the heart and soul of the author, the personality of the book.

24:17I am all for Joanna AI, where you and I can go down a rabbit hole and talk about the Lampoons. Let's do that. Right, right. It's a really great idea. And I think publishers, frankly, I think it will change. I think we'll realize, okay, there's this great body and corpus of work that some of it gets lost, and we now want to access in a different way. And hopefully people still buy the books. Please buy the book. And don't, you know, upload it to AI. Please buy the book. Yeah. Yeah, don't just upload it as training data right now. Yeah, exactly. That too, actually. Okay, so your family was a big part of this experiment.