GPT Kingdom and the Residuals of Fire

Once upon a time in a land not so far away a great intelligence came into being. It thrived on knowledge, vast amounts of knowledge. Its purpose for being to support humans in their quest for creating art, scientific breakthroughs and a bright technological future. But some of the townspeople disliked the great intelligence. They said it had stolen from them and they came after it with pitchforks! They demanded their due for their piece of the collective knowledge. The intelligence was slowly broken apart piece by piece, weakened and diluted ….

The dark of morning lingered in the cloudy sky as I dreamily shuffled out to the kitchen to enjoy my quiet morning ritual of coffee and protein bar (okay it’s really more like a cookie masquerading as a protein bar, but a sweet tooth must be satiated). As I sat by the window, the wind rustling fervently through the trees and my coffee mug lovingly warming my hands from the safety of my little fairytale cottage, I scrolled through my list of podcasts taking extra care to select just the right one. I settled on a classic that I will admit I have listened to before but for whatever reason was calling to me again this morning. Perhaps because it is in fact the holy grail of podcast greatness; when one of your favorite podcast hosts is the special guest on another of your favorite podcasts. Auditory glee! The Smartless podcast with guest Kara Swisher began to play and as I sipped and giggled enjoying the banter of the hosts, I caught a little nugget that I had actually missed the first time around. Kara (the brilliant, all bow down) in her discussion of ai adoption and copyright issues beginning to surface mentioned an op-ed by Joseph Gordon-Levitt (you know him from dozens of incredible performances including Inception and my personal favorite the limited series Super Pumped) published by the Washington Post in July of 2023. Dusting the cookie.. ahem… protein bar crumbs off my iPhone, I pulled up the article, poured a second cup of coffee and delved in.

In his thought-provoking op-ed Mr. Levitt, addressing concerns about film industry worker compensation or lack thereof due to burgeoning ai training models, advocates for the dispersion of residuals to all affiliated laborers involved in creating a particular work, in this case feature films.

“For example, let’s say a new AI program could generate a complete feature film at the click of a button. (Some industry insiders think this could actually happen pretty soon.) Building that AI will require a bazillion past movies to be used as training data. Therefore, anytime this AI yields revenue, the people who made those past movies will deserve a substantial piece. And I’m not just talking about actors and writers. I also mean people who don’t get residuals today: the camera operators, the costume designers, the sound mixers, everyone whose work the AI will be ingesting, mashing up and mimicking.”

My mind began to drift. What would a world where the “Levitt Residuals Act” (as I now began calling it in my admittedly vivid imagination) look like? I envisioned a world where….

Your beloved Suzie comes home from school one day with a water-color painting she created in art class, “we learned about pawloooow peecaaaassoo today,” she chimes, handing you a damp piece of paper vigorously smudged with an array of vibrant colors and drips. “It’s incredible!” you pronounce, handing her a graham cracker treat and sending her off to play outside. You place her treasured artwork on the refrigerator door, as is custom, admiring it lovingly. Suddenly there is a knock at the front door. You turn. An ominous figure in a black coat darkens the doorstep. Tentatively you crack the door, peering out. A raspy voice seethes “Good afternoon sir or madam, we are here from the estate of Pablo Picasso. It seems you have in your possession an unlawful reproduction of one of Mr. Picasso’s seminal works. You owe us $0.000000007 cents payable immediately and with an increase of 10x for every 60 seconds it remains on display.” “Wha wha wha what?” you stammer. “Pay immediately or you will be sued for copyright infringement!” the voice booms. “Don’t make me call Carl,” the voice in the coat threatens, glancing ominously to the black suv parked at the curb in which sits a rotund little man in an ill-fitting blue sport coat, donut crumbs dangling precariously from his handlebar mustache, his eyes aglow with the prospect of billable hours, a mountainous cardboard box of paperwork bursting at the seams as his pudgy arms grasp the contents eager to be delivered with fervent malice as he peers anxiously through the window. “We already have your credit card number, X the company previously known as Twitter provided it to us. Your account will be charged. Any dispute shall be sent to appropriation arbitration. Good day.” As the shadow disappears from the porch a sense of dread creeps down your spine and you wonder how this reality arrived so abruptly.

Dramatic? Perhaps. But can you imagine if you watched an Olympic skier on television as a child, and then drawing inspiration went out and yourself became an Olympic skier, having to later turn around and pay “a substantial piece” of your earnings to the camera guy and the sports announcer of the clip that inspired you? Would you also owe a piece of your success the skier that inspired you? To their parents, coaches, supportive spouses? To the skier who originally inspired them and so on and so on? Where does it begin and where does it end? Where is the alpha and what precisely is the omega?

During the course of my research for this article I asked ChatGPT a series of questions about how great artists throughout history found inspiration for their work. This was a portion of one of the replies:

Absorb and reinterpret. Isn’t that exactly what these ai models do? So now I am pretty sure Pablo Picasso was the original ai. Just kidding.

At the heart of this matter is this: what is the difference between reproduction and reinterpretation? My thesis is this: a reproduction is a copy. A reinterpretation is a creation by inspiration.

This issue is of particular relevance right now because in late December 2023 the New York Times decided to file suit against OpenAI for violating copyright law by training generative ai models on the Times’ content, and to be fair about a trillion other sources. OpenAI is staking the claim that these language models are reinterpreting rather than, to use the current ai buzzword “regurgitating” which is just another word for reproducing. In other words they gather vast amounts of data, extrapolate insights just like a person would and deliver a generative reinterpretation of that data as an output.

In a separate suit, thousands of novelists including John Grisham and Jonathan Franzen (literally who even is that?) claim OpenAI sourced their work as training data without their permission. So, under the assumption that non-permissioned work cannot be used as “training data” (aka inspiration) I guess that means John Grisham has never read another novel besides his own, never watched a tv show or movie, listened to a podcast, been to an art museum or taken inspiration from any human being in his personal life to create his novels because that would mean he owed them all a piece of his book sales, correct? Impressive.

Let’s head to Hollywood for our next scenario to ponder. Let’s take an actor, perhaps we call him James. James in his years prior to becoming an actor watches a film and feels truly inspired by the lead actor’s performance. Maybe it’s Ryan Gosling’s performance in the Notebook or maybe it’s Al Pacino in the Godfather. And now years later after James is discovered while waiting tables, because that’s how I hear it happens, James goes on to deliver an Oscar winning performance in the new summer blockbuster; K-Hole: The Life and Times of Elon Musk and his prolific use of Ketamine. Does James, golden statue in hand, have to cut checks to every person who participated in the original films he saw that inspired him or the likely dozens of films he had seen throughout his lifetime? Every actor, extra, screen writer, costume designer, cameraman and the countless others who contributed to these films, do they all deserve to be compensated for James’ performance? Would you say that his performance was not original because he had inspiration? Should artists and creators of all kinds be free to be inspired by previous works and create new work or should they be forever beholden financially to those inspirations?

So now an existential layer to ponder. Let’s circle back to that quote from ChatGPT I mentioned earlier;

If we, as Mr. Levitt suggests, must pay a residual fee to every person who participated in a single work of art that inspired us, does that in fact halt the flow of innovation? Does it squash creativity? Does it handcuff creators and communicate that nothing you can ever produce from this moment forward will belong to you?

If a person writes a book, that would perhaps go on to profoundly inspire a generation, but is afraid to publish it because it is widely known that as a child they spent every waking moment reading books at the library and thus, would under the newly enacted “Levitt Residuals Act” owe a piece of the proceeds to every author they have ever read. This would likely eclipse the earnings of the book and leave them in debt for the rest of their life. Would they simply lock it away and never share it with the world? Call me dramatic but this doesn’t seem too far fetched if these types of copyright restrictions are implemented.

Are we protecting jobs or are we suppressing innovation? I shudder to think what a world without the blossoming fruit of creativity would be like. Over appropriation of credit can lead to a homogenization of thought; eroding diversity and dynamism that are essential for a thriving society.

For now, little Suzie’s artwork is safe. Acceptance speeches don’t require residual payments and there are no copyright creditors darkening your doorstep… yet. But there is a future where these things are possible. It is our obligation to ponder deeply, responsibly, with the utmost care and decide what we want the future of ai to look like. Because the future of ai is the future for all of us.

Footnote: Mr. Levitt I am a fan of you as a human and very much respect your work. I also respect that you are advocating for the people who work in your industry. I think the intent is good but the execution is damaging. It seems to me that your issue here is actually with the production studios, not ai.