By Rob Garner
Cover photo by Robert Maxfield II
Additional photos by James Bland
Whenever the topic of the 1980s creative resurgence of Deep Ellum is written or talked about, names like Russell Hobbs, Jeff Liles, Jeff Swaney, and many others are all often rightfully mentioned in the same breath. One of the other figures who made a lasting impact on the scene is Dallas musician and tech futurist Tim Sanders. As a musician with The Affirmative, Cairo, and Code 4, and also as booking agent at the Theatre Gallery and Prophet Bar in the 80s, Sanders forged a mindset among musicians that took him well beyond Deep Ellum alone, and into Silicon Valley and Cambridge, Massachusetts. He bridged his Deep Ellum experience to become a successful tech executive, Harvard business fellow, author, and in-demand keynote speaker. For the last 14 years, he has been focused on the topic of AI, and how it will transform our world. As we have seen the hockey-stick arc of AI adoption in the last few years, he offers a window of experience to help business leaders and musicians know what to expect – and how to plan – for this new technological world we are entering into at an accelerating pace.
His DIY style, and ability to bring a communal ethos to the burgeoning Deep Ellum music and art scene was well internalized amongst many musicians of the time. Sanders espoused a volunteer and compassionate ethos, essentially along the lines of giving to get back. This philosophy was later core to early Internet and network culture, back when there was a real feeling that the Internet could change the world for the better. Though also congruently and independently espoused by several other leading network thinkers at the time, his ideas preceded the popular adoption of the Internet by many years.
In my own digital marketing book I wrote for Wiley/Sybex back in 2013, I had espoused a similar philosophy that had become somewhat prevalent amongst large segments of the digital community. But it was only after we concluded the interview that I realized that Sanders was the first person I ever heard speak of it conceptually, and that it had woven into a lot of what I do now – in both work and in life. In hindsight, I should have included him in my acknowledgements and credits, but I’m doing that right now.
Another fortuitous Deep Ellum happenstance for Sanders was when he went to work with Mark Cuban in 1997 and AudioNet, which later became Broadcast.com, and was sold to Yahoo in 1998. Cuban was a supporter of Sanders philosophy, and promoted his ideas within Yahoo! After the acquisition, Sanders was promoted to Chief Solutions Officer, and he took some of those same principles to the company, and Silicon Valley at large. Keep in mind that at the time, Yahoo! Was the #1 Internet media company in the U.S., and was as comparable in size and impact as Facebook, Instagram, or TikTok are today, in terms of audience share alone.
Sanders is also the author of many different books, including Love Is The Killer App, The Likeability Factor, Dealstorming, Today We Are Rich, and Saving the World at Work. He is a sought-after keynote speaker, and has been featured on on the cover of Fast Company. His books are a refreshing take on business strategy that are designed to not only help you improve business goals, but to also feel good about the work being done.
Sanders recent AI work has also landed him a role as Executive Fellow at Harvard, working with their Digital Data Design Institute. With his extensive experience into both the music world, and into the new world of AI, Sanders talks about all things AI – how to think about it, how to use it, how to benefit from its inevitable transformational impact on society and humanity, and “how we do music” as a whole.
Tim Sanders author profile at Amazon.com
Tim Sanders – LinkedIn
https://www.linkedin.com/in/sanderssays
Rob: Let’s first talk about your experience in music and the Dallas music scene.
Tim: So I moved here in the mid to late 1980s and I came here in a broken down school bus. Our band at the time, The Affirmative, which was a rock reggae band, had a very difficult tour, which led us to all lose our apartments in Tucson. We were looking for a place to move to because Tucson was a very depressed market. And when we had a gig in Lubbock, Texas, at a warehouse, we saw a Dallas Morning news article about the Deep Ellum scene. It had a picture of the New Bohemians and we said, ‘that’s where we’re going to move.’ So we got in the bus, we drove, we parked in a bus RV stop, and we went in and tried to get jobs at various places. And I got a sales job right off I-35 in Walnut Hill. After a couple of paychecks we could afford to move into our first band house.
And that’s how we got started. And our first gig that we played in town was at The Theatre Gallery on a Monday night. What was really cool was a couple of the people from the New Bo’s, for whatever reason, were at that gig, not to see us. They were just hanging out. It was a weekday night and they were in that famous cage inside the Theatre Gallery where you could drink because they had all the underage shows. So if you were inside the area, you could drink if you weren’t. That was our first gig. We played around sporadically because, of course, we didn’t have a following and we tried to build one up, make some friends. I met Russell Hobbs and he and I hit it off, not surprisingly, and he gave me an opportunity to do some work at The Prophet Bar. It was like odd job work.
And then eventually right before he went to Europe for his trip, he said, “you can take over the booking for it.” And that was really intimidating because the great Jeff Liles had been the booking agent there. So I knew I was not going to be him, but I did work on local acts mostly during the weekdays and then obviously promoters and others would bring in what roadshow played at The Prophet.
So that’s kind of how I got started in the scene and then had an opportunity to produce the Change Your Life Festival, which was really, really fun. And then as the years went by, when The Affirmative stopped being active, a couple of years later we started another band, Code Four, which was an industrial rock band. And that’s what we did for most of the time in Deep Ellum. And that’s where we made all of our great friendships.
Rob: Talk about the Change Your Life Festival. That’s where I know you from, at the 500 Cafe and all the planning sessions.
Tim: This goes back to my relationship with Russell. So when Russell went to Europe, he told me, ‘I want you to plan a festival. It’s going to be called Change Your Life, get 30 or 40 bands bands on the lineup. We’re going to put a bunch of money together with this festival so that bands are always going to have a place to play.’ But he also said to make sure that we go get a permit for it.
He told me about Officer Johnson, and what to watch out for because he [Johnson] was going to be skeptical about it and the kind of work I’d need to do. And he said, ‘oh, and by the way, if you could find a way to get some radio station involved, that’d make a really big difference to it actually working. And I said, “what is my budget?” And he goes, ‘just make it happen.’
And then he split and went to Europe and I started to do a bunch of meetings and we got the festival planning process going. There was a surprisingly large community of people who cared about having a festival in Deep Ellum to showcase it. And then the first breakthrough happened when Officer Johnson gave us a permit to shut down Commerce street for the festival, and that made a big difference at its scale.
But the big break came via Pete Tarantino, he was Redbeard’s assistant at Q102, and I met him through a friend of a friend. He got very excited about Change Your Live as well as some of the bands that were going to play there like Shallow Reign, which were really buzzing at the time. He convinced Redbeard to not only start promoting the Change of Life Festival, [but also] to play a few local artists as a lead up to it. And when Redbeard got involved, it became a much bigger event, drawing thousands of people each day. It was a special weekend in Deep Ellum, that’s for sure. And then we did a second one the next year, but it was on Lower Greenville, if you remember.
Rob: Yes, I remember that. I remember you giving these talks in the 500 Cafe and it was a very positive experience. You were letting everybody know they were volunteers, and that this was a collective kind of thing. There was a really good vibe. And this kind of segues into some of your books. With Love is the Killer App and your other books, you’re the same guy that was giving that talk to people there. Your books still share some of the same sentiments. You were pulling people together, you were motivating, you were doing it from a position of ‘let’s do this right and feel good about it.’ Can you talk about that a little bit?
Tim: Absolutely. And when I met you on Change Your Life II, we were making sure that everyone knew that they were going to be an equal partner. They were going to get paid. This is what bands wanted this time around, so we had a completely different model. And when I finally had the opportunity of a lifetime in 1997 when I went to work for Mark Cuban startup AudioNet, which became Broadcast.com, I was able to actually put this into practice at a much larger scale to go out to talk to people about really big investments in our services as well as webcasting really big events. I produced the Victoria’s Secret Fashion Show in 1998. And along the way I realized that professionally generosity was a breakthrough way to build a business or a career, for that matter. So I brought that ethos of every time you do business with somebody, make ’em smarter, share your network with them, but most importantly, be compassionate and be accountable for what it means to them as a human being. That really was how I was operating as a person, and that’s what Love Is The Killer App was about. And once you write that book, you’re kind of committed to that persona. So I’ve tried to expand upon that in the 20 years since then.
Rob: That type of apporach was different from other things going on in the music business at the time. In many ways, it was still exploitative.
Tim: The music industry as I know it was exactly opposite of Love Is the Killer App. And in the years after I went to work for Mark and then Yahoo! and all the other stuff, I’ve seen even uglier people in business than record people. But I wanted to change all that and that’s when I had the opportunity to write Love Is The Killer App, which is its own story. When literary agent Jan Miller discovered via in Dallas after a speaking engagement, she decided to sign and develop me because she thought there would be a big book in my perspective. She’d also developed Dr. Stephen Covey, Tony Robbins and Dr. Phil, so I believed she knew what she was doing.
Rob: You mentioned earlier that when you got to Yahoo! that they noticed you were coming from a different place. Can you talk about how they elevated it within the company, and how that philosphy impacted Silicon Valley.
Tim: Right after Yahoo acquired Broadcast.com, my manager Stan Woodward told Jerry Yang and Anil Singh, ‘you should get to know Tim because he’s very successful with our most difficult clients. They ‘love him’ – and there’s something there for your sales team or for something that might be valuable. One of their executives came to one of our meetings where I spoke to the entire sales team about something we were working on, and he was like, ‘oh my gosh, we need to bring Tim out to California to speak at one of our meetings.’ So literally a month after they bought Broadcast.com, I went to Silicon Valley, gave a talk at a big sales conference for like 600 of them, and the game changed.
So I started to develop this little training program, which was the basis of Love Is The Killer App. That actual training program was those three steps. Share your knowledge, share your network, share your compassion. And I think the other thing I’d put on my metaphorical gravestone is that I wrote Love Is The Killer App. It’s been translated into 13 languages and I’ve traveled the world to lecture about its concepts.
Rob: I’ve often said that my music experience was like being in a shark tank and it over-prepared me for some of the business situations I would get into.
Tim: I always loved hiring musicians as a result. Again, people who were in original music, which is hard, because the thing about cover music is you solve a specific business problem and there’s native demand for your service. Original music, you’re not solving anybody’s needs. You’re imposing on people and their properties so you can play. You’ve really got to build something from that. So when I find people, they’re like, I was in original music, we traveled around in a van on the weekends, we released two records on our own. I’m like, when can you start? And they go, well, I don’t have a lot of experience. I tell them, you have the instincts. I can teach you the technical stuff related to this job. What I can’t teach you is how to get kicked in the gut three times and show up again on Monday. And you’ve obviously learned that if you traveled around on a van on the weekends to get gigs and sell your tapes in the back of the room.
Rob: So I want to come back to the originality aspect of that a little bit later, but can you talk about how you’ve gotten into AI now? What you’re doing with AI and specifically what you think about the big picture is right now for music and AI?
Tim: When I left Yahoo to write books after just a short period of time, I really wanted to get back into business. So I started consulting with companies around digital technology and I started out with an interest in digital advertising. That was my background with Yahoo. But then after cloud computing got started in 2006, I immediately realized, “Oh, the cloud thing’s going to change the economics of IT.” And then Software-as-a-service was close behind that, which I got behind right away. I began to consult with companies about how to leverage that. And then around 2010, I first started getting interested in machine learning, inspired by Google’s leverage of it in search and Amazon’s use of it for suggestions.
Now today, machine learning has become the Nickelback of AI. It’s kind of a private pleasure if you still think it works. Everybody wants to talk about generative AI or agents. But what captured my imagination in 2017 was that Google researchers released a paper called Attention is All You Need.
It introduced the transformer, which solved a big problem in natural language processing and made breakthrough products like OpenAI ChatGPT possible five years later. But for me, that paper opened my mind to the idea that much like you’d see on Star Trek, the interface to data would be natural language and everybody would learn how to leverage it. Mind blown! This breakthrough, generative AI, would help AI take a great leap forward. Machine learning makes simple predictions such as ‘turn right, that’s a dog, not a cat or that’s a fraudulent charge.’
Generative AI creates content, code and delivers narrative insights! I started to follow Open AI , back in 2022, and have been focused on AI innovations intensely ever since.
This is the focus in my current job. I’ve recently joined G2, I’m leading Research Insights for them. They’re the world’s largest software review platform.
I’ve gotten to know a lot of academics over the last few years, and they are enriching my perspectives on AI and helping me think more like an economist than just a technologist or entrepreneur. Last year I went to Harvard Business School to meet with Karim Lakhani, the author of Competing in the Age of AI. I gave him my point of view about where I think it’s going, and he said, “I’d like you to talk to our faculty about this, because it’s a different way of thinking.” Several months later he led an effort to get me an appointment to serve as an Executive Fellow at the Digital Data Design Institute at Harvard.
My goal with G2 as well as Harvard’s research institute is simple: Democratize AI for business people of all types. Help them leverage its potential to improve their businesses and create opportunities for others.
If anybody’s just getting started in artificial intelligence, go read the book Prediction Machines, and you’ll really understand the basis of it. The authors say, artificial intelligence is a prediction machine. It takes information that you have and produces information that you don’t have. Up until recently, the predictions were very simple. They were binary. They were a series of zeros and ones you’d say, is this spot on this X-ray cancer? It says, no. Is this credit card charge fraudulent? It says yes.
So the predictions were very simple because the information that we had was very simple. But what happened with Open AI is they went out and scraped the Internet and then built a model trained on it, powered by the most powerful computers you and I have ever put our hands on. And people could now write instead of a search query like a keyword, they could write entire prompts as if giving an intern instructions. And you didn’t get simple predictions: You got solutions.
Well, that just changed everything. So now these prediction machines aren’t just saying zeros and ones anymore. These prediction machines are saying, ‘this is a book, this is a picture, this is a song, this is a video, this is a financial report’. So the predictions can now be longer and deeper, but they are still predictions. And I think that’s what’s important.
The other thing that makes AI transformative is that for the first time ever, we’ve decoupled prediction from judgment. In the past, both of those things happened at the same time in the mind of an expert, based on their educational background or years of experience. All of which took time and money.
Now think about that as a musician. An A&R person predicted whether a band was going to be successful, and then they passed judgment as to whether an advance at $100,000 would ever make their money back. So they were an expert. They had prediction and judgment in one human persona, and they were expensive. And if they were like Guy Oseary at Maverick and they got a couple of hits, they put a zero at the end of their salary, but the problem is that they became obsolete before they even knew it, or worse, aren’t loyal to their employer when another opportunity comes along
Now AI makes the prediction, but the human passes judgment. And what’s interesting, and they write about this in Prediction Machines, when the cost of prediction goes down, the value of human judgment goes up. With AI, we are entering the Rick Rubin economy. You don’t need to know how to run a mixing board, tune EQs or master a record. But if you know what good sounds like and whether the song “will work,” you can produce Grammy winning albums.
Here’s an example of that economy where AI can create what I call The Uber Effect. If you want to become a London cab driver, first you have to go to school called The Knowledge, which can take up to four years to graduate. This is to learn all the turns, all the routes, everything on the grid, and how traffic patterns change during different times of the week. This way, London Cab provides a delightful service where the driver doesn’t get lost or have to pull over and consult a map.
That’s a prediction problem. To be a cab driver, you needed street expertise, so you could predict the best way to get your customer from point A to point B. But, that experience or training requirement was a bottleneck to having enough drivers. Even if you found the capital for more cabs, only the market could deliver you enough experts to man your fleet. And as a result, there’s been a growing wait time during rush hours. London Cab thought they had a logistics problem: Get more cars and drivers. But they didn’t. It was a prediction problem: Create Waze launched, which created an AI driven turn-by-turn map system on your phone, and now you don’t have to go to school for four years to learn the knowledge. Now, all you need to be a professional driver is access to a decent car, a driver’s licence and good judgment not to have your radio on to the wrong station if the customer wants you to be quiet, not to tailgate the driver in front of you, not to honk too much to be polite.
So instead of there being only 40,000 drivers for all of the UK like they had before Waze, there are a quarter million ride sharing drivers in London alone, and everyone can get around better as a result.
When I think about AI being the end of the reign of experts, democratizing opportunities for those with good judgement, doesn’t that feel like punk, where you could have played guitar for a few years, wrote some amazing songs from your heart and had a platinum record?
Punk didn’t replace highly trained musicians. It created an opportunity for people with good taste and instinct to find an audience who would solve their distribution problems with passionate word of mouth.
So naturally, I don’t see AI replacing recording artists with good judgement and a point of view that springs from their human experiences. I see AI as being a part of their workflow so they can reach their goals more directly.
Rob: You definitely led into the next question, ‘does this democratize a lot of different types of music creation’?
Tim: I do think it can take a person who may not have had the background in EQ and compression and help them fix a song that’s got a terribly rumbling bottom end that’s getting in the way of everything and a vocal that just doesn’t cut through. And in the past, you might’ve had to be able to afford an expert who spent years and years and years in the studio to solve that for you. There are now artificial intelligence solutions for that like Izotope or even in DAWs such as Ableton and Logic. They allow you to focus on writing and playing your music instead of tweaking or fixing it. It doesn’t write or perform the songs for you, it enables you to bring your music to the world faster and in better technical shape. Even if you don’t have the cash to hire experts along the way.
I’ve never written a song using AI. I’ve never asked AI to write a chorus for a chorus. I have asked ChatGPT to bring down chord progression patterns for artists I’m studying, such as Portishead or more recently the XX. It helps open my mind as to their true writing style and acts as a teacher of sorts.
Here’s what you need to know: AI produces consistent results, but it will never produce remarkable results on a creative basis. This is especially true with generative AI, which is programmed to find an average solution to your prompt. So you can produce an image mid-journey that looks like what you’ve described in general and might even have great detail and vibrant colors, but you’re never going to produce a highly creative image like a professional human designer can make for you based on consultation and iterations.. So AI can help you do quick generic results, but it can’t help you produce results that will live after you do. I’m not blown away by apps like SunoAI or Stable that can make a song in response to a text prompt. Those songs are just awful. They are literal, generic and banal. So will that disrupt the jingle business? Maybe. Will they create the next Nirvana? Not on your life.
But do we think the radio’s going to be flooded with AI generated hits, either created by people with text prompts or by machines? Nope. It won’t resonate enough to achieve that spread effect great songs do. In fact, I think we’re going to even value more music that sounds and feels human and real, just like we value vinyl because it sounds warmer and more authentic to our ears.
Rob: With the majority of people I speak with about AI, there’s an overwhelming sense of fear, uncertainty, and doubt. I get it at a level of fear, however, but AI and generative AI is not new, right?
Tim: AI has been around for 50 years. It’s had a bunch of ups and downs, but this breakthrough around this paper, Attention Is All You Need was a game changer. It enabled us to talk to data with natural language. Because what makes generative AI different from everything else is that instead of being 10% faster at something, it is capable of being 10 x more efficient. So what we’re talking about is exponential.
So that’s why I think we’re threatened by it. We’ve never seen a technology that grew this fast. We’ve never been able to put something in a search box and have it write a damn book right before our eyes. That’s terrifying. But it doesn’t have to be right? Because I’ve now had conversations with a lot of artists and producers to really understand how a working musician says it’s making my life better. And there’s value for us if we make ourselves available to it.
Rob: In a scenario where AI is taking formulaic music and recreating it, is it possible to push people to be more original in their work? Will originality be more rewarded?
Tim: AI has not really changed the landscape for an original music artist today. You’re not competing with AI for record deals. You’re not competing with AI for record sales. Where I see it as a replacement currently is for jingles or for soundtrack work, or advertisements. One of the things I’m beginning to read a lot about is the idea that human beings can detect AI. So the performance rate of these AI generated articles is actually lower than a human written one because of the minor signals that AI conveys that screams “the generic answer.” And so I believe what we’ll see is a lot of content generated that just doesn’t work as well as when the humans generated it. I’m even seeing it in the business world where marketers declare, ‘we are going to start using generative AI to write articles and create our illustrations because it’s efficient.’ And then they pull back a few months later because all their metrics fell off. They are learning that while AI can accelerate processes, it’s not a creative solution.
For now, for the original musician, AI is not a threat to them. There’s a saying these days that ‘AI isn’t going to take your job, but a talented person using AI will. So it’s not that AI is going to take your music. It’s that another musician who’s learned how to use AI to generate great targeting lists, to grow their Instagram to 20,000 followers in six months that are engaged with them so that they’re putting 500 people in a room and you’re only getting a hundred in a room because you’re passing out flyers. That’s where we are really going to feel the impact of AI, not at our art level, but at all the other stuff we have to do as a band to succeed. Think of AI as helping you offload the no-joy part of being a musician, so you can focus on making and playing great music and earning a living from it.
Rob: Where there are formulas in music, let’s say Nashville songwriting formulas, perhaps certain chord progressions, or it’s been said the most popular chord progression, 1, 5, 6, 4, it transfers over genres. Does that get crushed a little bit? We can get AI to start our 1, 5, 6 4 progression in 10 different ways, and as we start to go do our thing on it and put our stamp on it, we maybe make that the clay. We’re going to build a sculpture from there.
Tim: That’s it. Now, you’ve made a good analogy there. So can AI give you the mother for a sauce, right? The starter for an idea. Can it give you the beginning that you tend to add on? What the early research is suggesting is that the original musicians playing genres like EDM may see valuable outputs from AI because trends drive formulas which drive an artist’s success in that situation. But with genres that reward originality, like alternative music indie rock, jazz and hip hop, it’s a different situation altogether. Those genres, it’s like putting lipstick on a pig. So you start with some shitty generic mother in your sauce, guess what your sauce tastes like? It tastes like Nickelback, right?
Why do we hate Nickelback? Because Nickelback originated on nothing new. The singer was very good, so they got a lot of fans in Canada. That’s why Drake got popular. Same thing. Not that he sang well, but he did just generic stuff and built it up. Yeah, I’m team Kendrick 100%. He’s an original article. I wouldn’t want to be Nickelback as my legacy.
And if your music doesn’t start through serendipity, they kind of hear it in the end result. Now, I’ve actually seen research on this where in the genre of jazz, an AI prompt engineer might say, “We’re going to feed you a bunch of Coltrane and you’re going to write a song that style of Coltrane.” And then they test those AI outputs on real jazz listeners four out of five immediately say, ‘I hate that one. I don’t know why.’ That’s called the Turing test, where we detect that we are interacting machine and the result just doesn’t feel right. AI music has not passed that test, and for that matter, neither has AI video.
So I don’t think the application today is to use AI to begin the creative process. I think where AI becomes important is to help finish the creative process. And this is what I’m starting to research now. Can AI help us do things in music that used to be cost or time prohibitive?
And I’ll give you two classic examples. So I was at an AI conference where they interviewed Derek Ali, famous for producing Kendrick Lamar. He’s Dr. Dre’s protege out in Los Angeles. And they asked him, ‘How do you use AI?’ And he said that it’s very limited for him, even though he’s learned how to use all the different tools to stay ahead of the game. He revealed that there’s a real premium on authenticity in the world of hip hop, so he’d never try to commercially release a beat he just made with AI.
So, he went on to explain that where AI is valuable to him is if one of his artists, like BAS, is out on the road and Sirius XM needs a clean version of one of his songs and they need it tomorrow, he can make it with one of these AI apps that can take Bas’s vocal and replicate a line without a swear word in it that sounds just like Bas.
I recently interviewed a producer who found another way to use AI: ‘Sometimes you might want to have something like a group of kids singing the chorus with you, example: Jay-Z and Hard Knock Life.’
That’s a very expensive sample element to record. It’s also logistically a nightmare. You’ve got to get releases, you have to organize it. And then of course, direct the kids to your desired results. It is just hard, hard to do a thing like that, right? So that’s very different, now with AI. In that case, AI’s not creating the song, it’s enabling you to provide the texture you are looking for to complete the sound.
I recently tried this approach myself where I took a snippet of recorded spoken word I’d gathered, and then prompted the AI to train on his voice, then create an extra line to the bit that came to me after our recording session. It passed the Turing test for me and now I’ll go back to that person who recorded the original sample and get a release to use his voice to create this synthetic line.
In this use case, artificial intelligence helps you deliver creative ideas as a musician quicker. So do six kids created by AI sound like six kids organized through a local choir? No, not exactly. But, if I never could have afforded that to begin with, and if it’s material to the song to have that good thing, then it’s adding value to my work.
All that being said, I still think for musicians, using AI for promoting their work is the best use case that balances tech savvy with authenticity.
Rob: Yeah. There’s so many branches we could go from here. Let’s talk about digital rights, because you get into some of that with samples and things like that. The one thing that strikes me as a lot of artists are saying, ‘Hey, some of my works are being used to train AI. Could be a chord progression, could be my style, my vocal style, and I want to protect my digital rights in this area’ where, frankly, there’s no precedent to my knowledge.
Tim: This is a very good topic. The first time I heard this whole topic about protecting ourselves from becoming dog food for AI machines was a podcast interview from a modular synthetic podcast called Why We Bleep. In late 2000, electronic artist BT was on the pod and when asked about this he said, ‘what I’m worried about with artificial intelligence is how it might use our work to feed their training sets.’ And then the host was like, ‘what do you mean?’ And he goes, “Listen, as an artist, you are the sum total of every creative decision you make in the process of music.” So from what chords you choose, to how you chose to hit the strings, to how you set the guitar levels, to how high the gain was when you recorded it, to all the mixing decisions you made, to everything down to the mastering, the whole chain of work is all a bunch of decisions you or you and your team made. That’s known in AI as a training set that can instruct a machine how to be you.
Great songs don’t fall out of the sky, they come from your unique process of production. As Ed Catmull, president of Pixar Entertainment famously says, ‘Toy Story wasn’t a breakthrough idea. It was a thousand problems solved.’ So a hit record is a thousand problems solved. So what BT is saying is what AI steals are our decisions, and they put ’em into a training set. And so when they’re training AI, what they’re really training to do is to think like Rob, to think like Rick. Think like Michael Jackson.
Rob: It’s not just about the creative work, this is about the process.
Tim: The whole process. So it’s not so much that they’re stealing your work, it’s worse than that. Your work is an output. That’s not your essence, okay? They’re stealing you because you are the sum total of every decision you make in the process of making music. In other words, AI is studying the logic behind music in order to write music on demand. And that’s what people need protection from. So the second part of the answer here, how do you protect yourself from such a thing? Well, you can cross your fingers that copyright will extend to training set someday, but I don’t know, because it looks like the big companies are going to write checks to the Reddits of the world, but they’re not going to ever write a check to Tim Sanders. So if there was ever a time where the value proposition of being signed by a label was never higher, it’s now, right?
Because the one thing I do know is the labels are already thinking about suing AI companies for stealing training sets for their artists. And mark my words, a year or two at the most from now, you will be reading about Open AI or Anthropic writing checks to Universal to eliminate any liability and have more access to those training sets. So there’s going to be a business model. Will it end up being a ‘pennies for your life’ model like Spotify? I don’t know. Or will it turn out more like a sync license model where an artist can make $90,000 with a placement? I don’t know. But I think if there’s ever been a time to have a publisher for your content, it’s now. Fortunately, there are so many flavors of record labels to choose from, depending on what resources you really need from them.
So as individual musicians, I don’t know how we protect our training set from getting out. But let me give you one idea here from a book I helped publish, titled Finding the Next Steve Jobs by Nolan Bushnell, co-founder of Atari. He was famously Steve Jobs’ first boss. So anyway, after Steve left Atari to co-found Apple, he stayed in contact with him over the years. And then right when Steve finally made his way back to Apple after being fired, went to France when Nolan Bushnell was having a launch party for Chuck E. Cheese, his second startup. The day after the party, Nolan and Steve are taking a walk along the Left Bank. And so Steve Jobs says to Nolan, ‘every time I have a good idea, somebody steals it. What the hell should I do about it?’ Do you know what Nolan’s answer was? ‘Be prolific.’
And that is the answer to dealing with piracy of all types: Be ‘prolific.’ People will always steal your good ideas, but they can’t produce your next great one. So I’ve lived that my whole life. That’s why I’ve written five books. I’m trying to write a body of work, be prolific. The one thing we can all do, is to continue to create content and put your twist on every new thing you do because the AI can’t keep up with you in that regard.
Rob: Getting back to the training set thing, I think you touched on something that has perhaps never been achievable by humans before. You’re talking about literally somebody’s process. Let’s just take an individual independent musician, and realistically, they might be playing music only five to 10% of the time, the other 90% or these other things about managing their business, about the recording process, about practicing, maintenance, upkeep, logistics, on and on. And there is inherently a process within that that enables the music to happen, which I think is what you’re speaking of. Are there any other comparable examples for an average person to have their training set copped in this way? Certainly not at this scale, but are there any other comparable ways where people have gotten these? We know, okay, there might be somebody that takes the time to turn their process and productize it in a way, make it to where it’s digestible and repeatable and that kind of thing. But what you’re talking about here is literally through AI, not just the content, the process itself in a way that another artist, let’s just say, and maybe this could be a good thing, right?
Tim: So let’s talk about that. So what you’re talking about here is a concept called digital twins. So digital twins is where you can make a digital version of you (or a machine, a city, a system, etc.). One example of digital twins is where you can construct a virtual copy of a city’s traffic system. I was just talking to a guy who’s working with us, he’s applying this concept in a project he’s designing for a city in Florida. With this city’s digital twin, as he redesigns traffic light systems to improve mobility and efficiency, he could run simulations a thousand times before he ever launches it into the wild. This will alert him to risks of collision and pressure test his models and math before he ever puts a citizen through the new process. And through the feedback you’ll get from the digital twin experience, your system design will improve. In this case, the twin becomes a digital mentor.
So there could be a way for your decision making process, say around something simple like songwriting, blues, and you’re going to mentor somebody who wants to be a blues musician, but he needs to start from zero from songwriting. So your essence of songwriting could be captured by artificial intelligence on demand, and there could be some sort of way for you to provide that to your mentee, much like when you mentor somebody and you sit down with the guitar and show ’em your magic. You could do it for money or for love – but it would be an interesting way to scale all of our goodness in the community.
So I think that’s a very positive use case, but let’s talk a little bit about the average person’s risk for their personal training sets to be taken. So here’s the good news. When you look at the artist’s work and ensuing training sets that are being procured (ahem, crawled) by like ChatGPT or Eleven Labs, those people have substantial bodies of work and commercial success that’s drawing the attention of the AI. It doesn’t have the space to listen to everything, it’s looking for a consensus of the right answer to feed their prediction models. It’s training on hits and widely talked about content. Same for books, by the way. Even though Love is the Killer App is best seller with hundreds of thousands of copies sold, I seriously doubt my training set for leading-with-love has been applied by any of the frontier models in their outputs. I’m just not commercially big enough to be interesting to their general needs.
So for most of us underground musicians, AI will never find us or it won’t find us enough times to decide the pattern should be encoded. So when AI trains on data, it’s really trading on these patterns of tokens. And if it sees it over and over again, and then it begins to see it linked to over and over again, it says, that’s significant. So kind of the good news there is that the very nature of AI commercially is that it won’t be mining the underground as much as it’s going to be mining it to find the guerrilla marketing and save-money-on-studios-or software hacks to get their music out.
Rob: When I hear some of the artists talk about being ripped off for training sets, I can’t help but think about this because ‘stealing’ has historically been most prevalent in music. Every musician will tell you, ‘well, I copped that lick from so-and-so,’ and Miles Davis said he stole this lick from Bird, and Prince said, ‘I stole this lick from Jimi Hendrix.’ Do you think there’s a conundrum there for some of these artists? At what point can somebody claim originality to the point that they’re able to essentially claim intellectual property rights without overstepping on somebody else’s work?
Tim: Yeah, that’s a really good question. So the answer I would give you is that I believe the reason we’re so sensitive now to Big Tech stealing from us is because they’re so darn wealthy. I mean, you look around, it’s like Nvidia is worth $3 trillion plus at this point. They’re worth more than all but five countries right now based on AI. It’s unfathomable, right? And so when you see that much wealth being created and you know that to some extent the world’s been scrapped for it, you feel like you should be getting paid a part of those gains. So I just want to say that for artists, I’m sympathetic to this because you see all this incredible wealth while the artist is just trying to buy another pack of strings. To think that their SoundClouds have potentially been scrapped by OpenAI is deeply upsetting and you can’t blame them. Now, does that mean they haven’t been stolen from in the past? Of course it’s happened, but many would say it’s all relative or that it’s actually an homage to greatness. As an old saying goes: Great artists steal.
But those people weren’t making trillions of dollars. Okay? So with what’s changed, it’s almost what we talked about earlier, what makes this different is the exponential nature of this, okay? Yeah. We’re not just creating a little wealth. We’re creating billionaires, thousands of billionaires. We’re going to create, and I’m not joking, a million millionaires in the next decade with AI. I’m telling you, as a starving artist, that should piss you off, right? So I’m sympathetic to it, but at the same time, it’s also easy for us now that we know that to disconnect a little from that energy and say, ‘okay, what can it do for me?’ I’m going to set fear and loathing aside and focus on what it can do for me. And I think that’s what I tell people all the time.
Look to leverage tech innovations like AI to accelerate your ability to build a mailing list to improve your draw or sell more records.
You could also use AI to build a target list of record companies, A&R people, musician-influencers and so on. This would help you operate more strategically to reach your commercial goals. And I’ll give you an example of something I just did this week. I needed to build a list of AI thought leaders for a project I was working on, and I needed a list of at least 50, but it needed to be a list where it was like, who’s the thought leader? Who are they with company or school? Give me one sentence to describe what their AI point of view is, and then what’s their social identity and what’s their email address?
And if you rank them on one, two, and three, like tier one, two, tier three, where they fit. So I can prioritize outreach and give me one little factoid about them that would prove that I’ve done my research. That’s a really hard list to build if you think about it. I built that on Chat GPT in less than an hour. Imagine a musician saying, ‘well, why couldn’t I use that to create a list of potential clubs in the south that had booked bands with a similar style, following as mine.’ You could train it on your website, YouTube, etc. That’s one place where I could focus my energy as a musician trying to book regional gigs, for example.
And none of those are shortcuts, as an artist. Those are cut-throughs as an entrepreneur. So that’s where I think we should direct our energy is in the part of our music business where the friction is the greatest and the money is the slightest. And I think that is where we can take free products like Chat GPT off the shelf or the new Claude and see how they can augment our work. You can do all these things. Just address the parts of your music project that need the most help, and you’ll be shocked at how much of a difference it’ll make.
When you hear about artificial general intelligence, when will it be as smart as people? And the answer is, it’s going to be a long time. So in 2024, I do not yet see where AI is going to help somebody write a better song per se. But I do see AI helping someone take a mix that’s a little muddy and make it so clear and simple that everybody can hear its true potential as a song.
One thing about AI is that it can make your work more discoverable. In the world we live in today, discoverability is the key to being successful. Can the person that you wrote this music for ever discover you? And it’s hard because there’s more artists and there’s more releases than ever. So discoverability is the lifeblood of whether you accomplish your goals as a musician, commercially, and this is where artificial intelligence can help you go find your audience, find ways to package your original music that can be found by the audience. Right?
So I was showing someone recently how they could use artificial intelligence to write their book description for their Amazon listing so that the average Amazon searcher for a romance novel of this genre in the south is going to find your book. Then they used AI to optimize the copy of their website as well as their LinkedIn profile.
The results of their sales went up 300% in the first week. They ended up going from selling something like a hundred units a week to something like 400 units a week because they became more discoverable and that led to a steady drip of new readers and buyers.
I think that’s what it can do for the musician: help us find out which clubs would like to play our music, which people would like to stream our music, which playlists could we try to get on, et cetera. I think that right now is the clear and present opportunity for musicians, and then it’ll get better from here.
There’s a great book out now called CoIntelligence by Ethan Mollick, where he writes, ‘today’s AI is the worst AI you’ll ever work with.’
Rob: Good point. That is how a lot of people are judging it now, when they say it doesn’t work.
Tim: It’s like when everyone said banner ads didn’t work in 1998. They were wrong as they improved to become paid links on Google or Facebook ads. In 2024, digital advertising is eating print, TV and radio’s lunch and sucking all the money in.
Rob: Last question. We know that the other fears of AI are creating alternate realities, creating fake people, networks, content, all those things. Is there a point that this can enable us to lose touch with reality? Or what is real? I should say they’re already doing it. AI is already being used as an excuse for what is possibly bad behavior. Is it going to question us all to know what is real?
Tim: There’s two ways to think about it: We’re going to have a harder time knowing what’s real in the future, but we’re going to value it. There will be movies that are made, starring synthetic people that have been created from stars of today. But will people be standing in line to go see the movie where the real George Clooney was actually in the movie? You know it. Having a real human in our entertainment experience will command a premium.
And I think those of us who consider ourselves very human and very original are going to have a one-of-a-kind opportunity to stand out in this weird Blade Runner looking future that’s in front of us.
—
