Transcript: The Challenges of AI-Generated Content

Copyright & Technology Conference 2023 Preview

Interview with Bill Rosenblatt

For podcast release Monday, July 31, 2023

KENNEALLY: Once upon a time, the agenda for a copyright conference listed only lawyers as speakers. In 2023, the lawyers have made way for software engineers.

Welcome to CCC’s podcast series. I’m Christopher Kenneally for Velocity of Content.

While artificial intelligence and machine learning aren’t new concepts in copyright or technology, they are the top draws for the 2023 edition of the Copyright and Technology conference, returning to Fordham Law School in New York City on Thursday, September 14th. Conference organizer Bill Rosenblatt has created a program that captures the copyright moment in a single day of panel discussions focused on AI, with bonus points for including the general counsel at OpenAI, the home of ChatGPT. Bill Rosenblatt joins me now from New York City with a preview. Welcome back to the program, Bill.

ROSENBLATT: Thanks so much for having me, Chris. It’s always great to be here.

KENNEALLY: We look forward to getting a look inside the program that’s coming back to Fordham Law School in September. Keynote speaker at this year’s Copyright and Technology conference is Jason Kwon, general counsel for Sam Altman’s OpenAI. Before he went to law school, Kwon worked as a software engineer. I have to ask, Bill, is his career trajectory now obligatory for anyone looking to make a name in either industry or law?

ROSENBLATT: Well, there’s been a history of people with technical undergrad degrees going to law school, but the vast majority have gone into patent law. I work with a number of such people when I do work as an expert witness in patent litigation. But those people going into fields that involve copyright – a little less common, but certainly more and more common for people with tech undergrad and law degrees to be doing law work for high-tech companies. There have been other examples of copyright lawyers who have had serious undergrad STEM or computer science training, although I would say we need more people like that. There’s not enough even now. So we’re certainly looking forward to having Jason Kwon. He’s eager to engage with the creator community on copyright issues, and I’m sure there’s going to be a great amount of discussion at the conference.

KENNEALLY: Certainly, generative AI is the subject of the moment, Bill Rosenblatt, in copyright and in technology. Tools like ChatGPT can compose essays and music, as well as create images and videos. One conference session, on what is an author and can AI be one, takes up the legal and intellectual challenge of deciding whether machine-made works are eligible for copyright.

ROSENBLATT: That’s right. Well, first, I want to mention that our AI-focused lineup is part of a whole year of programming that the Copyright Society is doing on AI and copyright. This is just part of an entire year-long program series that we’re doing, which I’ve helped to plan.

And this particular panel that you talked about on AI and authorship, which I’m moderating, has to do with to what extent AI-generated content qualifies for copyright in the eyes of the Copyright Office, which is ultimately the arbiter of who gets copyright on what. We’re going to be exploring two sides of the same coin. One is can a human author get a copyright on a work that’s been generated with hep from AI? And then conversely, can an AI itself be an author? Can an AI be established as an author?

There are two litigations whose lawyers are going to be represented on this panel, although they’re not going to discuss the pending litigations. One litigation is by a man named Steven Thaler, who is an inventor who created an AI that produced material that he’s trying to get copyright on on behalf of the AI as author. The Copyright Office rejected that, and that’s under litigation right now. And then the other is the representative of an artist named Kris Kashtanova, who created a comic book with the help of Midjourney AI, and she’s trying to establish herself as the author, even though AI helped generate the work. So there’s a lot of discussion about how much human involvement is necessary, how much AI is too much AI, etc. That’s really going to be a big question that we’re going to explore on this panel and hear about for years to come, I would think.

KENNEALLY: Bill Rosenblatt, organizer of the Copyright and Technology conference coming to Fordham Law School – certainly, determining whether published works are made by humans or created by machines is going to be more and more important as the volume of AI-generated content grows. A conference session on identifying and detecting AI-generated content tackles why we need to know the difference.

ROSENBLATT: That’s right. So we are having a panel on that, and there’s sort of two different approaches to this. One is for content that’s generated with AI to identify itself as such through metadata or a watermark – a watermark being one technique that was discussed recently in the meetings with President Biden among AI vendors, I think earlier this week or last week. And the other is trying to after the fact detect whether a piece of content was generated with AI in somewhat the same way as we now have technology to determine whether a file uploaded somewhere, such as to YouTube or Facebook, is a copyrighted work. There are these content recognition technologies that do that.

One of the panelists that Howie Singer, who’s the moderator of this panel, is going to have comes from a leading content recognition company called Pex, and she’s going to talk about how the technology that they have relates to this task of determining AI-generated content after the fact. And then there are others on that panel who have other technologies to describe, and then we’re going to hear about the legal aspects of that as well. So it should be really interesting. There are a lot of people in content industries who are very keenly interested in whether these technologies will work, how well they will work, etc.

KENNEALLY: That’s going to be important when we are confronted so much with misinformation and disinformation these days.

ROSENBLATT: Yes. And in fact, the technologies that are being brought to bear, other than content recognition for copyright purposes, have their roots in deepfake detection and plagiarism detection. It’s pretty routine now in colleges and universities – for example, I teach at NYU, and they do this at NYU – to run a piece of code on some paper that a student has written to see if it was plagiarized. You can obviously see how that could be applied to detection of AI-generated content. In fact, some of the vendors of that type of solution are trying to kind of branch out into AI detection. But we’ll see how well that works. It’ll be interesting.

KENNEALLY: September is certainly going to be an interesting month for you – an exciting month for you, Bill Rosenblatt. In addition to the Copyright and Technology conference, September will see publication by Oxford University Press of Key Changes: The Ten Times Technology Transformed the Music Industry, which you cowrote with Howie Singer, who played a leading role in the transition to digital music delivery at Warner Music Group. We will have to read the book to learn more about the other nine times, but what for you is the most interesting of these 10 technology-driven transformations?

ROSENBLATT: Well, first of all, the book is coming out in September. Copies will be available at the conference. I will be signing them for those who want that. But it’s funny that you ask that particular question, because the way we sort of pitched this book to publishers is there have been various books written – or papers, articles, written already about how the music industry was humming along fine until Napster came along and blew it up in the 1990s. The main thesis of our book is Napster, yes, blew up the music industry, you could say, but there were many other points at which a disruptive technology also blew up the music industry. We discuss in the book how that happened and what changes were wrought as a result of those technologies.

My favorite examples are – we know that Napster and P2P file sharing, internet file sharing, or however you want to characterize it caused a gross disruption of the music industry and a drastic reduction in revenue that the record labels were able to make – cut it by more than half. But there were actually two other times in the 20th century when pretty much the same thing happened that nobody really talks about anymore.

One was after World War II, when Columbia Records introduced the 33 RPM long-playing record, and RCA decided to go its own way and introduce the 7-inch 45 RPM record at around the same time, 1948-49. This caused a lot of confusion and frustration among consumers who were used to just one format, the 78 RPM record that had been standardized since the late 1920s. Everyone just said, oh, forget it, I’m listening to radio. In the middle of the postwar economic boom, record industry sales plummeted by more than 25%. That’s interesting. And they recovered in the mid- to late-’50s with the advent of rock and roll and the idea that the 7-inch had its own niche as a pop single kind of thing, whereas the LP was good for classical and Broadway and so on.

The other one that I love to talk about is around the the late 1970s, when we had the economic shocks due to the oil crises, the Iran-Iraq war, the first Persian Gulf war, which caused the price of petroleum to skyrocket, and petroleum is an input good to making records. So the prices of vinyl went up, and at the same time, the economy went into recession and people had less discretionary spending money. That was one factor. And then two other factors came together at the same time to create a perfect storm for the music industry. Factor number two was home taping. It was possible by then to make cassette copies of albums with very good quality on equipment that didn’t cost all that much money. I had some of that equipment in the late ’70s when I was a high school student. And third factor was simply that a lot of the biggest stars in pop music did not record albums in 1979. The Who, Fleetwood Mac, I think the Eagles – anyway, it’s a very long list of top pop stars who for one reason or another didn’t record anything. So the industry plummeted in value by over a third and did not recover until the mid-1980s, when the CD format started to come in and take off.

This is the kind of thing that we got to explore in writing this book, which was a lot of fun. And we’re certainly looking forward to the book coming out and to the reaction that it gets from the public.

KENNEALLY: Bill Rosenblatt, organizer of the Copyright and Technology conference, returning to Fordham Law School in New York City, and coauthor of Key Changes: The Ten Times Technology Transformed the Music Industry, thanks so much for speaking with me today.

ROSENBLATT: Thank you so much, Chris, and look forward to seeing you on September 14th.

KENNEALLY: That’s all for now. Our producer is Jeremy Brieske of Burst Marketing. You can subscribe to the program wherever you go for podcasts, and please do follow us on Twitter and on Facebook. You can also find Velocity of Content on YouTube as part of the CCC channel. I’m Christopher Kenneally. Thanks for listening.