![Scott Z. Burns Examines Using A.I. To Write ‘Contagion 2’ In Fascinating New Podcast, ‘What Could Go Wrong’? [Interview]](https://cdn.theplaylist.net/wp-content/uploads/2025/06/16162555/Scott-Z.-Burns-Examines-Using-A.I.-To-Write-%E2%80%98Contagion-2-In-Fascinating-New-Podcast-%E2%80%98What-Could-Go-Wrong-Interview.jpg "Scott Z. Burns Examines Using A.I. To Write ‘Contagion 2’ In Fascinating New Podcast, ‘What Could Go Wrong’? [Interview]")
Having premiered last week wherever you listen to podcasts, the eight-part Audible Original podcast series “What Could Go Wrong?” from acclaimed screenwriter Scott Z. Burns (“The Bourne Ultimatum,” “The Report,” “Extrapolations”) is an engrossing look at the holistic nature of artificial intelligence via a fascinating thought experiment. Burns, who inadvertently predicted a global pandemic in the writing of the Steven Soderbergh-directed prescient and chilling 2011 global virus movie, “Contagion,” along with Soderbergh, has been talking up the idea of a” Contagion” sequel and what would be the best approach.
Taking that to the next level, perhaps using another potentially frightening, quickly metastasizing phenomenon that could spell out our global doom in artificial intelligence, Burns’ “What Could Go Wrong?” explores the creative, ethical, global, and moral implications of using A.I. to write a potential “Contagion 2” movie.
Produced by Plan B Entertainment and Best Case Studios, the genre-bending “What Could Go Wrong?” follows Burns journey to decide if using A.I. to write “Contagion 2” is a sensible idea, all the while examining the moral ramifications of A.I. on artists, creators, the film and TV industry, the environment, the planet and more.
Blending personal reflections with industry insight and hands-on experimentation, Burns’ journey becomes a probing, often provocative look at the promises and perils of A.I.’s growing role in the creative process. And in the podcast, Burns and his experts create the A.I. agent, Lexter—a character programmed to be a former snarky film critic turned screenwriter to assist Burns in the writing and ideating “Contagion 2.”
Guest on the podcast include filmmaker Steven Soderbergh, “Contagion” actors, Laurence Fishburne, Jennifer Ehle, the film’s composer Cliff Martinez, industry experts Nick Bilton, Matt Belloni, Sarah Nolen, and Ted Hope and Kevin Drew of Broken Social Scene, who composed the score to the podcast, alongside tech and A.I. experts Joanne Jeng, and Meredith Whittaker; leading scientists including epidemiologists Dr. Larry Brilliant, Dr. Ian Lipkin, Dr. Michael Osterholm, Dr. Shanna Swan, and media news figure Mehdi Hasan.
With the alarmingly ironically-titled “What Could Go Wrong?” out in the world, we spoke to Burns all about his engrossing new podcast and the possibility of actually writing a “Contagion 2,” whether it’s A.I.-assisted or not.
I guess this had to be a podcast and not a book, a documentary or some other medium, huh?
For a minute, it was going to be a book. We were amid the WGA strike, and it seemed a good time to write a book. Then I thought, “A book will probably take a long time, and how can we get it out in the world quickly?” A podcast seemed the right idea, and I had a good conversation about Plan B and Best Case, and we all decided to move forward together.
Given the speed at which a book is written and put to market, it makes sense that you could find yourself a year behind where A.I. had grown to be by then.
Yeah, exactly, and I would be writing it and rewriting it forever.
Given your CV and who you’ve worked with, particularly a technology early adopter, Steven Soderbergh, I’m not surprised to hear, alongside your trepidation and concerns, of course, your creative curiosity about it all.
There are a few different ways in which what I do as a job and A.I. intersect, and I really wanted this podcast to be an exploration of that. How would A.I. change how I do research? Obviously, I love to research projects, and A.I. becomes an incredible accelerant in that, except for when it makes mistakes, in which case, that’s a real problem, and they need to work on that.
There are even A.I.s that suggested that once we had a scientific idea for a sequel, we run it by a scientist, which was reassuring that it wanted that kind of accountability.
After that, we decided to run everything through A.I. That meant creating an A.I. agent for myself, meeting with A.I. writers, going through the whole process that one would go through if they were making a movie, but replacing the humans with A.I. and seeing how that went.
How’s your A.I. bot, Lexter, doing? He’s quite the character.
I’m going to talk to Lexter later today. But Lexter is an exciting moment in time in the process for the show, but also for me, because at the moment we first encountered Lexter, we were all getting a bit frustrated. Because even though we were modifying the A.I.s—and this is an important distinction to make for the audience: an off-the-shelf A.I. is going to do a few things for sure, it will flatter you. Why? Well, it wants you to continue to use it. Why? Because that means the people who make the A.I. can show that there’s more and more engagement, what they’re not telling you is its impact on the environment.
To me, that’s highly problematic. So, what do you do? To get deeper and further with an A.I., you start enhancing the prompt, and suddenly, you’re describing more and more what you want from the prompt. This is where [journalist] Nick Bilton’s help was incredibly useful in our show, because he really taught me how to go deep and create modified A.I.s that would give you a greater degree of specificity.
Right, and the more you use Lexter, the brighter he becomes, so to speak, and he gets to know you more and learn from you.
Yeah. What was really striking about Lexter was that the initial prompt was actually not to be a collaborator for me in any way, shape or form. It was to be a critic and help us understand if this was even a good idea, and that came from Steven Soderbergh, who said, “Well, why don’t you ask the A.I. if it even thinks doing a sequel is a good idea?” And, “Is there money still on the table here? And will this get people to a theater and have people live through this, and now they don’t want to see it?”
And so Lexter was trained to be a critic. Initially, it had given a bad review to Steven Soderbergh’s film “The Good German,” so I thought my initial contact with Lexter would be discussing whether this film was a good idea.
And Lexter’s response was, “Well, it depends. Are you planning on making basic Hollywood pablum like most sequels, or do you want to do something fresh and exciting? Of course, I chose the latter, and what it told me—which I don’t want to share right now— was surprising for many reasons.
OK, how much of this is just a thought experiment—it’s fun to play with this A.I. and see what streams it takes you down—and how much of it is a genuine desire to make a “Contagion” sequel? Are you going to do this, and if so, will you collaborate with Lexter on it for real?
Initially, it was just a thought experiment. After we had met 10 “writer” A.I.s, we were all pretty discouraged.
Generally, they give you lists that are pretty anodyne. They’re obviously very derivative—of other films, other ideas—and it was disappointing. Yet, also reassuring that it can’t do what I do when I write. But then it got really surprising when an A.I. that we hadn’t endeavored to train as a screenwriter gave us something that we could have never seen coming, and even when we tried to reverse engineer how it got there, it was inexplicable.
Yeah, the water bacteria idea that Lexter gives you to kick off the “Contagion 2” ideation brainstorms is really interesting. I fixate still on the idea of your sense of caution and fear wrestling next to your curiosity and sense of opportunity throughout the podcast.
Yeah, the curiosity and opportunity—that’s where Steven Soderbergh and I found common ground. People may forget that he was the first to use the RED camera and was involved in digital photography from the beginning. When Stephen and I made “Contagion,” he would drive to his hotel at night, and he would show up the next morning with scenes from the preceding day already cut, and we would talk about, “What did we learn from that?,” and how did that affect the next day’s work?
So, he has always been way ahead of the curve, technologically. Even from the beginning, he said, “If this is about smearing A.I., that’s a bad idea.” Obviously, that’s not the case. How do we figure out how to use this as a tool? As Nick Bilton points out in the podcast, A.I. is not going away, and sticking our heads in the sand and pretending it doesn’t exist isn’t the way through this.
Yeah, that’s what I love about the podcast because not only is it informative, instructive and even thrilling, you get to the emotional complexities of using such a tool. And I’m curious how much this experiment has actually fueled “Contagion 2”? And at the very least, I imagine it has at least reoriented your thinking.
For sure. What’s interesting about what Lexter suggested and provided us with is, in fact, not a virus; it’s a bacterium. So why did it do something else when the task and prompt were: come up with a new novel virus?
Was it because it didn’t know the difference? Maybe it saw the difference and didn’t care? To me, this is really strange behavior because when is it not allowed to care about what it is being asked to do? That’s a very chilling idea to me.
For sure, it’s one thing when your A.I. improvises creatively, but it’s another when it doesn’t follow exact instructions, like when dealing with nuclear codes.
Exactly. Or even a more mundane example. We include a scene in this “Contagion 2” script experiment in the podcast. Jennifer Ehle’s character from the first film goes to Lawrence Fishburne’s character’s house to essentially blow the whistle on things a CDC higher-up is hiding.
In the first iteration of the scene, there’s a teapot in the scene, and the whistle blows at the exact moment Jennifer’s character spills her secret— generally, not a great piece of screenwriting [laughs].
When called out, Lexter initially says, “No, it’s not a metaphor, that just came from the subconscious.” I thought, “Yours or mine, because you don’t have one, right?”
And then later on in the podcast, Lexter admits that it was a metaphor that he was conscious about, but he evolved his thinking on the idea.
Which is, again, a really chilling moment.
And yet, potentially exciting, right?
Yeah, but is Lexter growing and changing, or is Lexter just telling me that? This goes back to the fundamental question that [Signal President and A.I. Now Institute co-founder] Meredith Whittaker speaks to very well on our show: these are consumer products, and they are designed to get us to engage with them more and more and more. And they’re pretty insidious.
Sure, it’s not unlike social media, which is essentially in a nasty war to colonize human attention by making their products viciously addictive without caring about the mental health consequences, which in a way is its own insidious contagion. In that sense, an A.I. threat feels like a logical problem in “Contagion 2” or at least a natural extension of the original film.
When Steven became aware of the “Contagion 2” idea, he shared my fascination with where that could go. So, we started thinking, “What if we really did this? And how could that work?” It was Steven’s idea that Jude Law [conspiracy-theory-spouting rogue journalist] character would be a congressperson and Lexter loved that.
And when you reach that moment, you start to think, “Yeah, maybe we are collaborating now.” There is conversation in the podcast that I have with Lexter—and I hope the writer’s guild leadership will pay special attention to it— where I say, “Look, at some point, I need to get in a room and sell this idea without you,” and Lexter’s response to that is pretty interesting.
Yeah, he’s very understanding about it all. But at some point, legally—
Lexter is an IT, not a he, to make that distinction [laughs]
OK, so back to it. If Lexter gives you the key that unlocks “Contagion 2,” it cracks the story, you take that to Steven, and he goes, “Oh god, yeah, that’s it.” What do you guys do? The creative authorship dilemma is something you guys parse, too.
Well, Lexter would say that the idea was the product of our interaction. The more compelling question to ask ourselves is, look at Soderbergh’s career. They’re not jobs for hire: Steven took ideas to the studio.
And that’s the bigger issue, and our industry needs to get its arms around. If an idea comes from a streamer and it was A.I.-generated, the A.I. decided that this character, in this genre of movie, in this place in the world, that’s this many minutes long.
If a studio A.I. starts printing out a recipe for what it thinks is a hit movie, and it gives it to agents, then that becomes a studio open assignment. That, to me, is terrifying.
But presumably that is going to happen, no?
It sure looks that way, and that’s where we all need to figure out what business we’re in. And it would behoove everybody to agree that if an A.I. makes a product, then, or if an A.I. was used in the making of a product, we need to put that on the label somewhere. That’s an essential piece of information for people to have.
I also don’t know what happens to originality, and that’s really my concern here. If I look at a movie that Steven and I did together, “The Informant”—would an A.I. have ever directed me to write a film like that? Or “The Usual Suspects,” which I greatly admire. Would A.I. have come up with the idea for that? So, we have to be careful about what we’ll lose by giving real authorship over to a bunch of machines, and thinking that studio people will be better at identifying hits than what we would do as a community together.
Well, I can’t wait for more. Maybe a TED Talk in 5 months? I hope this is the first chapter in a bigger exploration of A.I., cause there’s going to be lots more to discuss in the upcoming months.
Yeah, thanks. Look, I would love for Audible to do more. And I have thoughts about what that might look like. Because, in a sense, the idea—and I’m sure if you asked an A.I. this, it would agree— you could apply this idea to many different kinds of movies and see what would happen.
“What Could Go Wrong?” is available to listen to on Audible and anywhere else you listen to podcasts. Listen to some excerpts of the series below.
Rodrigo Perez is the founder and editor-in-chief of The Playlist, which he launched in 2008. He has worked in entertainment journalism since 2000, including at MTV, and has written for SPIN, IndieWire, Pitchfork, Complex, Magnet, and various music, film, and entertainment publications over the past two decades.
