Alex Kantrowitz, journalist and author of Big Technology, joins Tom and Vivek in the studio to discuss his road to journalism, ad tech, and the business and ethical considerations of generative AI. Alex recounts his career shift and discusses the impact of generative AI on journalism, drawing from an incident where AI tools were used to plagiarize his work. The discussion also covers the role of AI in improving performance across fields like journalism and software engineering, and its applications in law and music. AI can simply automate specific tasks, but is it likely to replace the necessity for critical thinking and domain expertise?
Alex Kantrowitz, journalist and author of Big Technology, joins Tom and Vivek in the studio to discuss his road to journalism, ad tech, and the business and ethical considerations of generative AI.
Alex recounts his career shift and discusses the impact of generative AI on journalism, drawing from an incident where AI tools were used to plagiarize his work. The discussion also covers the role of AI in improving performance across fields like journalism and software engineering, and its applications in law and music. AI can simply automate specific tasks, but is it likely to replace the necessity for critical thinking and domain expertise?
Listen to the episode and read the transcript at superset.com
***
Guest: Alex Kantrowitz
LinkedIn: https://www.linkedin.com/in/alexkantrowitz/
TW: https://twitter.com/Kantrowitz
Super{set} Twitter:@supersetstudio, @ClosedSeshPod
LinkedIn: https://www.linkedin.com/company/superset-studio/
Twitter: @tommychavez, @vsvaidya
Speaker 1:
Welcome to The Closed Session: How to Get Paid in Silicon Valley, with your host Tom Chavez and Vivek Vaidya.
Tom Chavez:
Welcome to season four of The Closed Session podcast. My name is Tom Chavez.
Vivek Vaidya:
And I'm Vivek Vaidya.
Tom Chavez:
V, this is the beginning of a new season, right?
Vivek Vaidya:
Is it?
Tom Chavez:
I think so.
Vivek Vaidya:
Okay. Well, you keep track. I don't.
Tom Chavez:
No, you're good at counting. The math thing has always bewildered and confused me.
Vivek Vaidya:
Yeah. I tried to do this, this season but I think I mixed up the episodes last time I was trying to do this, so-
Tom Chavez:
Right, and then we took the keys away.
Vivek Vaidya:
That's right.
Tom Chavez:
Yeah, you messed that up badly.
Vivek Vaidya:
That's right.
Tom Chavez:
So in this episode, we're going to be getting into the world of gen AI, which is hot, hot, hot. I can't pick up the paper these days without reading about it.
Vivek Vaidya:
Online or digital.
Tom Chavez:
Everywhere.
Vivek Vaidya:
Or print, I should say.
Tom Chavez:
That's right. And we're lucky to have a really thought-provoking, super engaged guest with us today to just chop it up and explore this rich, rich topic. So Alex Kantrowitz, please say hi.
Alex Kantrowitz:
Hello. Welcome. Oh, God, I'm so used to saying welcome.
Tom Chavez:
Well, because-
Alex Kantrowitz:
Great to be here with you guys.
Tom Chavez:
Because your podcast is way up in the stratosphere. Well now, we get to turn the tables, right? We get to pepper you-
Alex Kantrowitz:
That's right.
Tom Chavez:
... with questions. Well man, it's so great to have you. So Alex is a distinguished writer and journalist who extensively covers the tech industry. He rolls deep. And we also were just remarking before getting on the mic here today that we knew Alex, or were in similar spheres many years back, because Alex is a ad tech operator and hardcore veteran before you went into journalism and writing, right?
Alex Kantrowitz:
That's right. I always used to view, or still view, ad tech as the crimes and courts of the technology fields. When you want to learn how a society works, you go look at the crimes and courts first. Then you could think about policy and community-level involvement and reform. And having started off first as an ad buyer, actually I was in-house ad buyer for a couple years and then I moved into sales at Operative, I felt like that was really an unbelievable view into the underpinnings of what was then the digital economy, which was largely ad related. And coming out to San Francisco after that, I got a chance to cover companies like Facebook and Twitter and Snapchat and LinkedIn and Google and see the way that they operated with a focus on the ad business. And I felt like a lot of my counterparts were focused on the policy and that it was skipping a step.
Tom Chavez:
Right.
Alex Kantrowitz:
I think that when you know what's going on in the crimes and courts, that's when you really understand what's happening inside these companies. And that was very helpful to me.
Tom Chavez:
I mean, that resonates, Alex, because we've seen all the crimes and misdemeanors because we were in the boiler room ourselves with the first two projects, a company called Rapt and the second one called Krux. So crimes and misdemeanors. I don't know if it's leading to the courts yet, but that's maybe a topic we can kick around here a little bit.
I'm fascinated by the journey, right? Because I'm trying to think now of anybody who started as a buyer and then an operator and then a very celebrated journalist writer. How'd that happen?
Alex Kantrowitz:
Yeah, so it is a very untraditional path. Usually, what happens is journalists get into the field idealistic and then they say, "Hey, I want to make money." And then they go into ad tech. It happened in the opposite way for me. Not financially driven. I feel like if I would've stayed selling SaaS software for ad tech startups, might've been much better financial decision. But here's what's happening.
So I came out of school in the middle of the financial collapse. Had no job out of college, backed my way into an ad buyer position and then move from there to sales because I wanted to go from the periphery to the core. I wanted to be involved in the revenue engine, and sales was the way to do that. And ad tech, which is a native New York industry, felt like the right place to go. And this was like 2008, 2009, 2010, 2011. The world was changing so fast.
I mean, majored in labor relations, so it's not like I was coming out of school and saying, "I want to be a journalist." But I had written a little bit for the college paper, and just having that front row seat to the industry changing was fascinating to me because... And you guys, I'm sure, anyone who's been inside a company knows exactly what this is like, where you experience one thing and you see the press doing two things. Either A, ignoring it or B, getting it completely wrong. And that's the position I was in. And I was like, "Wait a second. Hey, I think I can write. I'm seeing the industry shifting. There's a trade press here that's completely missing what's going on. I'm going to throw my hat in the ring and start getting some pieces out there and share what I'm seeing."
And it was this amazing moment where Twitter was starting to become what it was. And instead of having the traditional path where you have a journalist who goes to the local paper and then you go work your way up to this state paper, the city paper and then maybe national, for me it was completely opposite, where I was working in industry, but I could tweet my stories, share them out on the social web and then they started picking up traction. And then all of a sudden, I was getting inquiries from the front door saying, "Hey, could you write this for us? It seems like you understand that ad tech thing." And the more I did it, the more I realized I'm spending way more time thinking about the next story I'm going to write than the next client I'm going to try to prospect into or close. Maybe instead of getting fired, I should try to go ahead and push forward on this journalism thing and see where it goes.
Vivek Vaidya:
Oh, so you were writing while you were still at Operative.
Alex Kantrowitz:
Well, you guys want to hear a funny story? I've never told this before, and I probably shouldn't.
Tom Chavez:
Okay, let's go.
Alex Kantrowitz:
I was writing basically from the moment I started as an ad buyer.
Tom Chavez:
Oh, wow.
Alex Kantrowitz:
Because... Yeah. I was working for New York City's Economic Development Corporation and I learned social media marketing. I just knew that we were going to get a better ROI on Facebook and on LinkedIn than we were doing print ads in the New York Real Estate Journal. There's no question about it. So I sat in for AdWords classes at NYU school Continuing Professional Studies, learned a little bit about LinkedIn and Facebook, and again saw these new platforms emerge, saw Tumblr emerge, saw what was going on with Twitter and... Oh yeah, and Meetup. The New York tech ecosystem was so interesting.
So I started writing basically right away. And I actually sent this email to a person I wanted to interview. And that person was connected in some way to my company, and I didn't identify myself as working at the company. And they forwarded the email to our comms-
Vivek Vaidya:
Oh, boy.
Tom Chavez:
[inaudible 00:07:01].
Alex Kantrowitz:
... saying, "This reporter has reached out to us." And I sent that email during business hours. So this is to anybody who's aspiring about thinking about going into journalism: don't do it this way. But I was within an inch of my life getting fired. They decided for some reason to keep me.
Then as I moved into the next job, which was at Operative, I had a conversation with them because I had to completely put all writing on pause after that incident. And I had this conversation with Operative and I was like, "Hey, can I write?" They're like, "As long as you're not writing about us, about our competitors or for our competitors, have at it." And that gave me the license to go ahead and start saying, "Okay, I had a little bit of success in the beginning there before I had to do what I needed to, to keep my job. How else can I start to poke around here?" And that's where things really started to gain momentum.
Tom Chavez:
Nice. Well, big ups to Mike Leo, the CEO of Operative, because I'm sure he had visibility into that. That's a really enlightened thing to do because so many companies are so focused on rules and sticking to useless... What's the harm, right? And now, look at what it's turned into. That's an awesome journey. Really great.
Alex Kantrowitz:
Yeah, it was very helpful. Very helpful to be able to do that, very helpful to be able to see inside these companies at first before doing the bigger stuff.
Tom Chavez:
No doubt.
Vivek Vaidya:
Yeah. So one of the things that's happening, and you've written about it too, is, as Tom was saying in the beginning, generative AI. And you've been affected somewhat negatively by it in terms of Petra at Rationalist trying to take some of your stuff and putting it out as their own. What was that experience like?
Alex Kantrowitz:
Yeah, this was the weirdest situation. So I had written this story basically right after New Year's about how the creator economy was a little bit overblown, and the story got some pickup and it ended up on Techmeme. And Techmeme will sometimes... For listeners who don't know Techmeme, it's a great aggregator of tech news that you can go to and basically see what's going on in the tech world every day. Highly recommended. I'm there every day.
And so the story gets picked up on Techmeme, and they list some of the stories that are adjacent to the main story underneath that headline. So there was this story from this subsect publication called The Rationalist that seemed to agree with what I was saying. Unfortunately, it didn't just degree with what I was saying, it plagiarized me, but did it in the most interesting way, which is this author took my text, ran it through a number of generative AI tools and had those tools effectively... You can tell ChatGPT or something like, "Rewrite this." Had the tools remix the content to make it look as if it was completely original and just basically coming to the same conclusions as mine and published it.
And I was reading it and was like, "Oh, this sounds interesting." And then I was like, "These clauses sound extremely familiar. I'm pretty sure I wrote them." And I was looking for the link and there was no link. And I went to the comments and I was like, "Plagiarism is ridiculous. You'll never get ahead this way," and posted. And as I'm looking through the comments, I see the author has replied to somebody else saying, "These are the generative AI tools that I used to put this together."
Vivek Vaidya:
Yeah.
Tom Chavez:
[inaudible 00:10:36].
Alex Kantrowitz:
And I was like, "Wait a second." So it's not that generative AI took my work and spit it out and someone took credit for it. It's that someone was able to use this tool, or these tools, to take what I had written and extremely easily remix it and post it as their own. And I never would've known if it wasn't for Techmeme. And by the way, their story made it to the front page of Hacker News, which is always good for thousands of views, and mine didn't. And it really annoys me to this day.
Tom Chavez:
No, it is irritating, right? And let's use that as a launch point into a slightly larger conversation about the implications here, right? Because journalists, artists, creators are first in line in terms of dealing with all the negative consequences potentially of gen AI. Actually, just this morning, in fact, I was talking to somebody who wanted me to... or emailing with somebody who wanted me to schedule something. And it's not just Calendly. He has an AI assistant. I know these AI assistants are out there, but they're getting really good. So made me think about, "Okay, how many of these jobs are going to get replaced, supplanted, laundered into a mix that nobody can discern later who did what?" Right?
So I'm wondering, Alex, if you might share perspective. Are you net positive, net negative? How are you thinking about the longer term implications? Because that's a very personal thing that just happened, right? Your name just got lost in the wash, the machine, and then somebody else goes and claims it as their own. Well, this is your livelihood, right? What do you make of that?
Alex Kantrowitz:
Yes. It's a great question, and I really do remain bullish that these tools are going to be a force for good. I really do think so. And I've been able to use some generative AI tools to really help the presentation of my newsletter, Big Technology. I mean, Midjourney has been a complete revelation to me. And this is kind of like the additive effect, where I'm a solo creator so I've been using, or I had been using, images from free stock photo or whatever it was. And let me tell you, the top of my posts did not look good, and I spent hours trying to find the right image there. And Midjourney is now illustrating all the stuff that I do.
And then in terms of generative AI, it's just... Right now, if you ask Bing or ChatGPT to write, it's garbage. It can write. It can do a form letter very well. But almost every story I create, I'll throw it in there and be like, "Do a better job of this." And there might be one or two sentences that it improves, like the definite clunky ones that an editor would rewrite, but almost everything else it makes worse.
And there's something to be said about this generative AI that it can take the bottom 50% and raise it up to average. So that's actually great news for the bottom 50%. It cannot take the top 50% and make it the top 1%.
Tom Chavez:
Right.
Alex Kantrowitz:
It's just not there yet. And we'll see if it'll ever get there. And that gives me a little bit of hope because I think that instead of replacing, it can help refine. And eventually, the value of someone that does this as a human is just going to go up.
And this isn't original, but I'm just going to share it anyway. We have the ability now to manufacture unbelievable pots and different types of artistry with machines. But what does everybody want? They want the handmade vase or the artist-drawn painting. And you sign that, and it's worth way more.
Vivek Vaidya:
Mm-hmm.
Tom Chavez:
Right.
Alex Kantrowitz:
They're so interested in original work, they almost paid for NFTs for five months.
Tom Chavez:
Well, that makes a lot of sense to me because I'm a stickler for good writing. And I'm not just saying it because you're on the podcast. You're a great writer. It makes total sense to me that the absolute best that ChatGPT can come up with is a poor man's approximation of one of your average paragraphs. However, we work with a lot of people who are not great writers. And so in that sense, it's great to get people who don't like writing to come up with more coherent... I mean, it drives me crazy when I read sentences that are ill formed and don't make any sense. So you're bringing up the average for all-
Alex Kantrowitz:
It's amazing.
Tom Chavez:
... [inaudible 00:15:08]. But then maybe, and the implication of what you're saying is, it ups the ante. It increases the level of play and the value of creators who can ride in that 0.01% at the top, which is maybe not a bad thing, right? I mean, it's obviously a good thing. Seems like.
Alex Kantrowitz:
Yeah. So this generative AI in some ways... I mean, it does so many different things, but part of what it is, is a communication technology. Midjourney and DALL-E allow you to express ideas through images. And something like ChatGPT and Bing Chat and Bard, with the text tools they can help you write better. In fact, Grammarly, which has been around forever, is an essential tool for anyone who's trying to write. It really does help. It taught me things about writing that I didn't know.
And I do think that ultimately, your value inside a company is going to be largely based on how you can communicate. And much of the communication in business in particular is done through writing. If you could bring that up through generative AI, that's great. Actually, it makes businesses run better, probably leads to more growth and more jobs. So I'm definitely not on the side of this is the end of times.
Vivek Vaidya:
Yeah. Alex, what you said about the bottom 50% getting up to average, but the top 50% not being able to get through to the top 1% with help of tools like ChatGPT resonates perfectly in the software arena as well. Because we are told all the time, "Oh, Copilot will just write all the code for you." And it barely helps a junior engineer become a somewhat capable standard software engineer because there's so much left that you still have to do as an engineer to put to code to actually work.
So one of the things that also comes up a lot is, are people going to lose their jobs because of generative AI, right? And like you were saying, I think what I heard you say was that jobs are going to become different, and you're going to have to learn how to use these tools to get better at your jobs. What do you think about all that?
Alex Kantrowitz:
Yeah. On average, I think that's correct that this will lead to growth and lead to more jobs. And on net, we're going to be better. But as with any technological disruption, there are going to be people that will lose their jobs without a doubt about it.
It's interesting. You meet two types of companies, or sometimes it's the same company, that says these two different messages out of two sides of its mouth. So one is the company that says, "We're going to use technology to do whatever we can to be more inventive." And that means basically simplifying processes to make more room for employees to invent. And you have other companies that say, "Yeah, any technological advantage we're going to have, we're going to use it to reduce full-time employees and just expand our margins and cut down our cost basis." And I do think that in any economy, you'll have companies doing both. And it won't be clean. It's not necessarily going to be clean.
Even my former employer, Buzzfeed, I'm not saying this is a direct relation, but they started doing more AI content and they closed the buzzfeed newsroom. Now, I'm not saying that the AI took the jobs, but I'm saying that if you're an executive, you can think about, "Well, maybe I can create more customized personalized content and I don't need the writers anymore."
Tom Chavez:
That's right.
Alex Kantrowitz:
I don't think that's a good idea, but I understand that there are going to be companies that do that. And we're also... One of the things that I'm curious what you guys think about is, well, this technology's going to get better.
The way I describe ChatGPT is how it exists right now. I can handle what's going on right now. Coders can handle what's going on right now with Copilot, but is there going to come a day where you're going to say, "Hey, write this for me," and it can do as good of a job as I do? Or Vivek, maybe they'll code the same way you can code. I've learned this year that saying that's never going to happen is not a really good stance.
Vivek Vaidya:
No, I-
Alex Kantrowitz:
There's always a probability, and I'm curious where that goes.
Vivek Vaidya:
No, I completely agree with you. I think my position, also, on the potential for this gendered AI technology to do amazing things is immense. I think that one, we're barely scratching the surface of this. I agree with you. I-
Alex Kantrowitz:
Yeah, can I give an-
Vivek Vaidya:
Yeah, go for it.
Alex Kantrowitz:
You go ahead.
Vivek Vaidya:
Go for it.
Alex Kantrowitz:
I was at the Collision conference in Toronto last year about a year ago. Or was it Web Summit? I think it was Web Summit in November. Oh, it was Web summit in November, right before ChatGPT launched. And I was on stage with Nick Thompson, who's the CEO of the Atlantic. And people can go find this on my YouTube page. You can watch this session.
And I was saying that... I asked all these questions that have now come up with ChatGPT. Who's responsible if somebody uses this and plagiarizes? And we were talking about different fields in media that might be automated because of it. And I mentioned that eventually, you might have a podcast host or a news anchor that could get automated. And Nick was like, "You don't really believe that, do you?" and all the crowd was on his side. But now, those questions are much less clear-cut as they were when we had that conversation. I mean, I sat on YouTube and I watched a solid half hour of the Joe Rogan podcast with the OpenAI CEO, Sam Altman, and both of them were AI. It was pretty compelling.
Tom Chavez:
Yeah.
Vivek Vaidya:
Mm-hmm.
Alex Kantrowitz:
So I think that just the speed at which this stuff is changing can really make our answers today look somewhat irrelevant or obsolete-
Tom Chavez:
Look, and it's been noted-
Alex Kantrowitz:
... a few months down the road, which is... It's very interesting and also somewhat concerning.
Tom Chavez:
But yeah, to your point, it's been noted that things are happening fast. It's hard for us because we're in the middle of it to adequately discern the velocity, right? Because to your point, Alex, no, no, no. After November, 2022, it's a whole new world. And that's just a couple of however many months it's been.
I remember when I first read The Singularity, my initial reaction was, "Okay, just a lot of bullshit. Come on. No." And then you read Kurzweil's argument, and then I'm also thinking about our day jobs, Vivek, where 10 years ago I think we're putting the numbers to this on a whiteboard in the office like, "Oh yeah, it's a hundred X improvement. It's a thousand X improvement in storage costs, performance ratios." No, it's more like a million, right?
Vivek Vaidya:
Mm-hmm.
Tom Chavez:
So we're seeing it in our own personal day jobs. And then you go to the balcony and you think about, "Okay, what is this?" Of course, we're going to have AI-enabled podcast hosts. We have all these tough problems already happening around is that machine generated or is that a human, right?
Vivek Vaidya:
Yeah. And I mean, why go far? One of our companies, Alex, we're actually writing AI to test software. It's going to generate tests based on the software you write. I didn't think that was possible five years ago.
And you asked a good question which is will I, just like you, describe your ideas to Stable Diffusion and it generates images for you? Midjourney will generate images for you. Will there be a world or time in the future where I can say, "I want to build a system that does X, Y and Z. I want it architected like so using these technology platforms with this kind of scalability requirements. Please go write all the code for me"?
Tom Chavez:
Mm-hmm.
Alex Kantrowitz:
Right. Yeah.
Vivek Vaidya:
And I think that's possible. I think that's possible. Is it possible for it to be 75% correct right now? Probably not. But 30% correct, 20% correct? I think so. Yeah.
Tom Chavez:
Yeah, and it's... Look, within a decade it's upon us. We're going to be gesturing, gesticulating at whiteboards. We're going to, "Well, I want to sub-system take some of this data and massages it, moves it over here." It's going to be happening. And I know we sound like we're high. So for people listening here, totally sober. I had a couple shots of caffeine as I usually do on... But no, I mean this is not Mundo Beyondo stuff anymore.
Hey, as a postscript to what you were talking about, Vivek, around engineers, just because I happened to be at a thing yesterday, I talked to a guy. This person said with great certainty that he has a 10X engineer who's freaked out because there's a junior engineer who's crawling up his tailpipe doing stuff with Copilot, and that the 10X engineer is saying, "Oh my God, this is unsettling me." My initial reaction was, "Okay, he's probably not a 10X engineer if that's happening today." But there's still a really important kernel of truth, which is that the ceiling is rising. It's upping the ante, to your earlier point. Whether it's journalism, whether it's code, whether it's software testing, we can go on and on and on. The entire level of play is getting raised, and the velocity at which is happening is still hard for us to comprehend.
Vivek Vaidya:
Yeah. I'm curious to get your perspective on this, Alex, because in the software world, as Tom was saying, my observation so far is that ChatGPT or generative AI will produce the code for you. But if there's something wrong, there's a bug in the code that it produces and doesn't work properly with the systems that you need to integrate it with, then that junior engineer just cannot do that, cannot fix that, right? So how does that manifest itself in other fields like journalism, for example, right? Will gen AI replace editors?
Alex Kantrowitz:
Yeah, I don't think it's going to replace editors. Certainly not. There's so much that an editor does that's more than just line editing. I mean, if AI could replace editors, then Grammarly would've made them obsolete. It's that good.
Vivek Vaidya:
Yeah.
Tom Chavez:
Yeah.
Alex Kantrowitz:
But that's just line editing and tightening in the story. It's not contextualization. For an editor, story judgment and direction is the most important thing.
Outside of the journalism field, it isn't seen because it's a hidden practice. But when an editor sits with a journalist and says, "What are you thinking about?" and the journalist just spews everything in their notebook. And the editor who's working across five different stories and has been doing this for a couple decades says, "Hey, that little thing that you mentioned, that sounds interesting to me. Why don't you dig into that?" The reporter goes out and finds it. That is journalism. That is the job. It has nothing to do with writing, actually.
In fact, writing is like the last mile of journalism. I'm a journalist. I mean, obviously I podcast and I write. I freaking hate writing. I mean, the worst days of my life are writing days. We know in this house when it's... I write on Thursday. The newsletter goes out Friday. Thursdays are like no-fly zones. It's just not a good time. And that's why this stuff is exciting, in some ways, to the field of journalism because if you can go out in the field, get the stories and then have the AI assist with the writing, that's actually very exciting, something that can actually make you a lot better. So I don't really see it as something that is going to make any editor afraid, except if they're handling copy that's been written by ChatGPT and they have to-
Vivek Vaidya:
Fix it.
Tom Chavez:
Right.
Alex Kantrowitz:
... make that gruel look appetizing. But it is going to come into other fields. And right now, there's been some very interesting applications.
So this is a case from late May, early June from Ars Technica. They wrote about this. So this is the beginning of the story. A lawyer is in trouble after admitting he used ChatGPT to help write court filings-
Vivek Vaidya:
Yep, [inaudible 00:27:41].
Alex Kantrowitz:
... that cited six non-existent cases invented by the artificial intelligence tool. And the person, his name is Steven Schwartz and he's of the firm Levidow, Levidow, & Oberman, which by the way is an absolutely amazing law firm name. It's like generative AI would pick Levidow, Levidow, & Oberman if you asked it to come up with a law firm name.
Tom Chavez:
With a good sense of humor too.
Alex Kantrowitz:
He says he greatly regrets having utilized generative artificial intelligence to supplement the legal research performed herein and will never do so in the future without absolute verification of its authenticity. So I think we're at this point where everybody's doing exactly what bottom-level writers are or bottom-level engineers are, which is saying, "Hey, how much better can this technology make me at my job? How much more productive can it make me?" And I'm going to say, there are some moments where it really, really can work. It can help in research. Like, use Bing Chat? My goodness. It can really help.
Vivek Vaidya:
Yeah.
Alex Kantrowitz:
But when you start to rely on it for that last mile, so to speak, you're screwed. This lawyer is going to be in... First of all, he is going to live in infamy forever for being the dummy that did this. But second of all, he's just living proof that in any profession... And you've given the coding example. So from right brain to left brain, you're just not going to have this technology necessarily be able to replace the core work of a function yet.
Vivek Vaidya:
Yeah. Actually, just building on that, we recently announced funding for one of our companies called Boombox, which is-
Alex Kantrowitz:
Nice.
Vivek Vaidya:
... a platform for musicians to collaborate and connect and earn money from their work. Think of it like GitHub for musicians. And as part of the product, we've just released, a tool called Boombot, which is essentially a generative AI-powered assistant that helps musicians create music. So you can say, "Give me a chord sequence for a new song I'm writing on climate change and in the style of U2. And it needs to be a high-energy song," or something like that. You can give it prompts like that. And we massage the prompts and then work with ChatGPT to get back chord sequences and help the musician write the song, even lyrics for that matter.
I mean, it's opened up questions for us around intellectual property. And like what we were discussing when we started this conversation is the question of intellectual property and creative rights of artists, journalists, et cetera. So how do you think about that? How do you think that technology can play a role in that?
Tom Chavez:
Right, and just to add to that because I, with our amazing team, I'll just ask village idiot questions like, "Guys, there has to be a way to sprinkle some bits on this music that one of our creators is coming up with on the platform such that we could watermark it somehow." So if it does get washed into the machine and a musician has the experience that you had, Alex, right?
Alex Kantrowitz:
Mm-hmm.
Tom Chavez:
Well, it can be tracked. It can be copyrights, and ownership can be enforced. We don't know if that's possible yet, but we're exploring it. But how do you think about, to add to what Vivek's getting at here, how should we be thinking about intellectual property and creative rights for journalists, content creators, everybody?
Alex Kantrowitz:
It's a great question because it does seem like throughout history, part of the creative process has been effectively stealing and improving, right?
Tom Chavez:
Yeah.
Alex Kantrowitz:
On my show, we talked a little bit about the Ed Sheeran court case where he got sued by Marvin Gaye's Estate for stealing part of his song for one of Ed Sheeran's great songs. And yes, those two little riffs could layer on top of each other fairly seamlessly, but did Ed Sheeran steal the song? Not really. And it is interesting now that yeah, of course machine learning can do the same and just do it at scale, but I think what's going to come, and we talk about this a lot, what's going to come into play is what sort of data sets are they training on? I guess there are going to be efforts to make it practically difficult to just pipe in all of Reddit and pipe in all of Twitter and train on it.
Imagine the lawyer. By the way, lawyer's a good case to talk about. That lawyer that used ChatGPT and started fabricating cases, eventually that lawyer is going to be able to use a system that will be trained on the actual corpus of our legal history and be able to cite real cases. That's happening, right?
Tom Chavez:
Mm-hmm.
Vivek Vaidya:
Mm-hmm.
Alex Kantrowitz:
And so the question is, then, how do we... When we build those models, the past history that we train on to get him the accurate case, there's going to be someone who's going to help work through the rights to that case law. You can't just pipe in a textbook. You need to work through the people who own this stuff. And that will actually be a part of the value chain. So this idea that you can just all purpose ChatGPT your way, at least right now, to success doesn't seem right. There's going to be specialized bots that are trained with specialized information, potentially using the base that we have now of the GPT models as a foundation for that conversation.
So I think that ultimately, we're going to actually start to be running into some of the walls here that will protect, in some ways, copyright owners. So that will happen. And in cases where if I have a competitor that uses ChatGPT to write a story that's similar to one that I'm writing about a similar topic, I don't really have the option to sit there and bellyache and be like, "Well, that's annoying. That shouldn't be legal." No. Welcome to the new era of competition. I now have to figure out how to beat the machine-enabled publication. And by the way, if I can't, put me out of business, because I have no reason to be doing what I'm doing if I can't beat that.
Vivek Vaidya:
Yeah. Because then it's back to your point about the bottom 50%, and the top, 50%, right? The bottom 50% just came up to being average with the help of AI and beat you. So you're average, too, if that starts to happen, right?
Alex Kantrowitz:
Yeah. Time for me to find something else to do.
Vivek Vaidya:
Yeah.
Tom Chavez:
Well, on that inspiring and still sobering note.
Alex Kantrowitz:
Wait, that was the final word? Geez.
Tom Chavez:
Let's go have a really stiff drink now.
Vivek Vaidya:
Yeah.
Tom Chavez:
But no, man. This was such a wide-ranging, interesting conversation. Alex, thanks a million. This was good. This was really great.
Vivek Vaidya:
Thank you, Alex.
Alex Kantrowitz:
Yeah, thanks.
Tom Chavez:
Until the next episode.
Vivek Vaidya:
All right. Don't forget to sign up for our newsletter. You can sign up at superset.com. Thank you, everyone.