Interviewed by Michael Laskow
Editor’s note: The first paragraph is an answer from Part 2 of this interview. We’ve included it this month to give context to the beginning of this part of the interview.
Adam Taylor: Adobe, Shutterstock, Pond5, and others have announced [that they’ve made deals to let AI companies harvest data from their catalogs] with pride. So, I think it’s hubris, and I don’t think there’s a really equitable way to do it. I think that if you take a defined set of music and you get X amount of dollars, and you say, “Okay, I’m gonna give X percent of that to those composers,” that one [the first] generation of people, then maybe that’s fair, trackable, but after that, what happens? All of that information, all of the learning that comes from that music is then no longer associated with that music, and there’s no way for the next generation of learning to be able to adequately or fairly compensate [the original creators]. And even if there is a mechanism for compensation, if you can’t measure it and track it, it’s never going to be fair. It’s not transparent, and I think we all know that systems of distribution of revenue or income or advantage are never fair and never even, and if you can’t track it, you’re not going to get your fair share. So, I think it is a false flag to say, “Yeah, we’re compensating composers for taking your music and allowing a computer to create new stuff.” It’s a false flag; it’s just to build their business and reduce their costs. That’s my opinion.
Michael Laskow: I agree with you, Adam. And to your point that needing an adequate way to track things, I’m a little dismayed that the PROs haven’t started using AI to track usages. I would love to see where you stand on that, because I’ve long been not a fan of the formulas [they currently use] to calculate payments and distribution of performance royalties. Audience members, would you please raise your hands if you love the formulas that the PROs use to track your music. There’s not a single hand up in a ballroom of around a thousand people.
But it seems like a really good use for AI. Rather than taking a sample of a primary market like New York City and saying, “Okay, you know if 800,000 people watch this episode of a show in New York City, that means in St. Louis it was this many, in Seattle it was that many,” and then they do some mysterious calculations and generate a payment to the creators. Technology and even AI could do that in a heartbeat; how do you feel about that use?
Well, firstly, I think the technology solutions to identify the performances of music in most cases already exists. I don’t think that it needed AI in order to do that. You know, the corollary question to the people in the audience that you just asked about how many people like or trust, or whatever, the formulas. How many people can explain them?
“I do believe that AI is going to play a very, very important role in improving the accuracy of the information that is available to the PROs so that they can come up with better and more equitable formulas for distribution, and that they will be more transparent.”
Right. Not a single hand went up for that either.
And truthfully, you know, I’m in the business and I can’t explain them in many cases. In some cases, I can, but not enough. I often tell people that I can go to any country in the world—any Third World country anywhere—and stick my ATM card in the machine, get my money out and have it get properly deducted from my account. The ability to do this exists, we live in the world of microtransactions, and I believe that it is incumbent upon the PROs to do all this. I don’t think it’s an AI issue, they should be using whatever tools that are available. I do know that organizations such as ASCAP, who have been very good… I know things that they are doing trying to advance their systems. They understand the issues and are working with large consulting companies and technology companies to really take a holistic look at the whole thing. And now that AI is here, how does that fit into it?
The formulas that the PROs use, the way that they do it, is complex. There are a lot of politics behind the scenes for all the PROs, there’s that issue as well. But I do believe that AI is going to play a very, very important role in improving the accuracy of the information that is available to the PROs so that they can come up with better and more equitable formulas for distribution, and that they will be more transparent. That’s the direction I believe they are going in.
There are societies, though, that are slow. One of the things that I often talk about, which is not an AI issue, is that PROs around the world—this is not an American issue, it’s the foreign PRO issue—have known about the Internet for some time and they should already have been able to figure out how a producer in any country of the world can acquire a worldwide performance license, rather than having to go to every single society around the world to get a performance license. It’s ridiculous; there should be a grid that you can plug into and say, “Okay, I want my license.” Then it’s up to the societies to figure out how to distribute the money, because that’s really the role of a society: to gather repertoire, collect performance revenues from people performing the music and figuring out how to distribute it. Right now, they do it within each country, but it needs to be global. We’re constantly getting asked for global performance rights, and it’s a very challenging thing to be able to figure out, and there isn’t an existing solution.
I agree! It’s also amazing to me that you can use your ATM card anywhere in the world and the transaction happens in seconds, yet collecting money as a writer takes a year and a half. I don’t understand why that is. Anyway, problems to be solved.
“We did a lot of extensive testing of the [AI search] system that we’re using, and it works pretty well. It’s not perfect, but it works pretty well.”
But one problem that I know is that you guys are using AI—not a problem so much as an improvement, I guess—and it’s also an area of AI that I am very, very familiar with. We’re not using it at TAXI, so don’t get scared anybody, but using AI for music searches.
You mentioned very early on in our conversation that you’ve got half a dozen people that are music directors. They need to know your incredibly vast catalog. They need to understand and interpret a request; they need to know your catalog; they need to know if they can go five or ten percent outside what the brief said, because something might serve the same purpose and is close enough. All those little human factors that go into it. I believe you guys have started using AI in some capacity to assist with those searches, which I think is great. Like I said, I am really familiar with that technology and have friends who have developed it at a very high level. I guess I’m looking for a consensus from you on this, that the way that works best is having human input, doing that interpretation on the inbound request and a human to check whatever it is that—for lack of a better way to say this—whatever AI spits out on the other end. What are your feelings about it? How is it working out for APM so far? Do you have the same kind of hope that I do that it could really speed up that process?
I do. On searching for music, I think it’s extremely valuable. Up to now we’ve used keywords, and we have algorithms that we’ve developed that we try to utilize those key words and those tags and the descriptions of tracks and things to be able ascertain what a person may be looking for and what are the best tracks. To do that we have a set of results.
We are using AI in a couple of ways on the search side. One of them is audio similarity. We use a system that you can take a track from our own library as a C-track. You can upload a track, you can grab a YouTube or URL or Vimeo or TikTok or, I think, Instagram, and you can put that into the search bar, and then it will deliver a set of “similar results.” There’s a gradient of similarity, and we cut it off at a certain point, and then we allow people to further refine their thoughts—you can add additional search terms. So, let’s say you find some kind of a track but you want one of a particular genre, or you want the beats per minute to match or you want something else like some instrumentation, and so we allow you to narrow it down.
We did a lot of extensive testing of the [AI search] system that we’re using, and it works pretty well. It’s not perfect, but it works pretty well. But I’d be able to further refine your searches then it gets even better. We also have a metadata similarity algorithm that we’ve used for years where you have a track as a C-track, and then you say, “Find me tracks that are tagged in a similar way,” and that finds a different set of tracks. It’s interesting and some people like that better, and some people like the audio similarities. None of these is perfect, nor are human beings perfect. It’s an additional tool.
The only thing that we’re testing right now and then going to be implementing soon is natural language processing. So, you can pipe in, “I’m running down the hall in the hospital feeling anxious but hopeful,” and hopefully it finds something that works. I think that we have a responsibility to try to deliver the best tracks to somebody but also give them a choice and a variety of ways of finding things so that they can use those various tools to hopefully get to a piece of music that they like.
I completely agree. You know, I’m personally excited about the future. I think that the future is gonna be bright, but so much of what our industry relies on is human intelligence, because there’s so much interpretation to be done. I love this industry, and I’m really, really hopeful about people finding good ways to use AI.
But I want to get that elephant out of the room now and talk about something else before we wrap this up, which is… I’m gonna give a rare endorsement from this stage. I’ve only done this maybe three or four times in 27 years. Adam is also the chairman of the PMA [the Production Music Association]. And frankly, in the early days of the PMA, I knew a lot of non-exclusive library owners, and some of them were heard to say, “The PMA’s raison d’être is to keep out the non-exclusive libraries.” So, I didn’t take it very seriously, or frankly, look into it all that much. But over the years the PMA has earned their stripes and earned a lot of respect. And I know a lot of people at the top of the PMA; I know a lot of people who are members of the PMA. So, I just want to encourage you guys in the audience, especially in the face of AI creeping into our industry, to join the PMA so that you can have another great resource for learning about this stuff as it develops.
TAXI and the PMA serve the same audience, and our purpose is the same, which is helping people make money doing what they love, which is creating music and making sure that they’re treated fairly.
Is there anything you’d like to add about the PMA, Adam?
Sure. I can add a few brief things. Firstly, we were founded in the mid-1990s, really before non-exclusive was a thing. I joined the PMA in 2001, so it was about six years into its existence. And the raison d’être was to gather people together to talk and really to try to make inroads with the PROs, in terms of them respecting and understanding more about performance, and about the music library industry. That was really the reason that it was founded in the beginning, but over the years it has morphed into something completely different.
The PMA has three mandates: It is to educate, to create community, and to protect the value of library music. That’s what we do, and I think we do a good job at it. We have virtual sessions; we have demo derbies and all kinds of things all year. In the fall we had the Production Music Conference. We had close to 900 people attending this year from all over the world, including tech companies, libraries, composers, lawyers and all manner of people who make a living from production music, or who would like to. There were a lot of composers there. It was really an exciting and wonderful event, and I agree with you 100% that we [TAXI and the PMA] should work more closely together. It doesn’t interfere or compete, we support each other.
So, for those who are interested in having a variety of different potential income streams by working in music—commercial releases and library music and studio work and playing out live, and whatever the other things are that a songwriter or a composer or anybody in the music industry can do. It’s incumbent upon people, I think, in order to be successful today to have a… especially to be able to give up your day job and just work in music, to have a variety of potential income streams. And library music can and in my opinion should be one of those, because it can be viable, and it’s a lot of fun. You are writing for a different purpose and you’re really stretching your brain in a new way.
So, I really encourage everybody in this room to sign up. It’s not expensive. I don’t remember the exact price, but it’s under $100 per year, I know that. So, if you go to PMAMusic.com all of the details are there. If you’re a member, you can participate in the online events—I believe they are free—and the Production Music Conference, of course, in late September or early October. It has a fee, but again, it’s not very expensive, and we would certainly welcome everybody who wants to come.
So that’s my pitch. I appreciate the endorsements, and I think, back at you. TAXI is doing an amazing job and is really, really important for songwriters and composers and anybody in the industry who may not know necessarily how to get their stuff out, and to have all of the tools at their disposal that they need in order to be successful at their passion. I commend you on that.
Well, thank you, Adam. And thank you so much for taking the time to do this. My schedule is crazy; I know that yours is even crazier. So, let’s make it a point to have dinner in the near future. Thank you again so much, especially for doing this while you’re out of town and making yourself available via Zoom. Ladies and gentlemen, Mr. Adam Taylor. [applause]