Yep, it’s going to be another one of Those Columns, but bear with me: there are some interesting, and maybe even not terrible, implications of the latest generation of AI tools for the geek culture industries of comics, boardgame and videogame design, entertainment and various aspects of the business.
Talking TED: I spent the last week in soggy Vancouver, BC attending TED2023 (home of the TED Talks), a confluence of mostly tech industry movers and shakers engaged in Very Serious Conversations about the direction of science, technology and society. Think of it as a comic con for smartypants, complete with panels, previews and an after-hours con bar cene. They even had Avengers, Fantastic Four and Frank Frazetta books in the bookstore, courtesy of sponsor Taschen.
AI (artificial intelligence) was, unsurprisingly, a big topic at the conference. Greg Brockman, the CEO of Open AI (makers of ChatGPT and Dall-E) was one of the speakers. So was Tom Graham, founder of a company called Metaphysic, who demonstrated a super-creepy tool for making realistic deepfakes for use in film and TV. There were also a few skeptics, including one speaker who advocated for enforcing an international ban on AI research by nuclear force if necessary because the danger to humanity was just that great.
I was hoping that folks at the top of the industry could clarify some of the uncertainties about where this is all going, or maybe reassure the worriers that there is some kind of a plan. Bad news: there is no plan. No one is driving this car. No one knows where it is going, although it is going there fast. And no one is giving too much thought to collateral damage for creative professionals.
Fortunately, there is some good-ish news.
AI and the Geek Culture Industry. If you’ve spent any time playing around with ChatGPT or one of the image creation tools, you’ve probably encountered a combination of emotions. This is cool! This is creepy! This is oddly frustrating! Why does the girl in that picture have seven fingers?
The dark secret of this generation of AI tools is that you still need some human skill to turn the output into professional-quality work. How long that will be the case remains to be seen, but for the moment, there’s an argument that AI can be a useful tool to help boost the output of people who already have some idea of what they’re doing.
For example, ChatGPT knows an awful lot about comics history and continuity. If you’re a writer for the Big 2, or any other company with a deep connected universe, it can be a great tool to find out what stories have already been told, and get clues to the backstory of obscure or disused characters, typically a process that can take writers days or weeks of prep time. Here’s a little chat I just had:
ROB: Has Rocket Racoon ever faced Thanos in the comics?
CHAT GPT: Yes. One notable instance was in “Annihilation: Conquest – Starlord #1-4” in 2007, a miniseries written by Keith Giffen and illustrated by Timothy Green II. In this story, Thanos is resurrected and seeks to acquire the powerful artifact known as the “Cosmic Cube.” Rocket and his fellow Guardians of the Galaxy join forces with Star-Lord, Ronan the Accuser, and other heroes to stop Thanos’ plan.
During the battle, Rocket and his team are able to retrieve the Cosmic Cube, but Thanos manages to steal it from them. Rocket then comes up with a plan to use a device called the “Kree Psycho-Prism” to overload the Cosmic Cube and destroy Thanos. The plan works, and Thanos is seemingly defeated.
Rocket has also faced Thanos in other comic book stories, such as “Infinity Gauntlet” and “Thanos Imperative.” In “Infinity Gauntlet”, Rocket is one of the many heroes who join forces with the Avengers and the cosmic entity known as Galactus to stop Thanos from wielding the power of the Infinity Gems. In “Thanos Imperative”, Rocket and the Guardians of the Galaxy team up with the cosmic superhero Nova to stop a cosmic threat that has taken over the Negative Zone, which includes the resurrection of Thanos.
Go ahead, try it. It’s like having your own personal Douglas Wolk (author of All the Marvels) or Peter Sanderson to sort out all the complicated bits of trivia and continuity.
From continuity to creativity. I have also used ChatGPT to work out story structures based on simple ideas, like “provide a synopsis of a hardboiled detective story involving corruption during the construction of the Empire State Building.” Working from that simple prompt, I was able to manipulate the chatbot like a holodeck, adding, deleting and modifying story elements, generating a complete outline framework, teasing out character arcs, filling in historical details, and creating a chapter outline. I do not write much fiction, but if I did, I think I could write a serviceable manuscript draft from that framework much, much faster than using a more traditional process.
At the very least, the chatbot automates quite a bit of the typing and retyping of structural elements like outlines and bullet points. If you have an idea of where you want to go with particular story ideas, or are working within certain story constraints or genre conventions, this can be a big help. Inventive game designers can probably figure out a way to use this for playtesting and other labor-intensive steps as well.
Currently, the great limitation of ChatGPT is that the prose it produces is completely without style or individual flair. Even when asked to imitate a noted stylist like Edgar Allen Poe or Hunter S. Thompson, the best it comes up with is cliched and uninspired.
But that’s good! It means that it is not coming for the jobs of actual creative writers for the time being. If you make your money on the power of your individual voice and your ability to breathe human insight and empathy into mechanically-constructed plots, not only do you not need to worry about getting replaced by a chatbot, you might even be able to increase your productivity significantly. And your readers might be able to enjoy more of the stuff that makes you you.
TL:DR, threat or menace? As a professional writer, I can see the advantages and limitations of the chatbots more clearly than on the art side, where, to me, the ability to conjure a gorgeously-rendered image out of nothing but a prompt feels like magic. But maybe artists who are fully in control of their craft are starting to see some of the benefits of human-AI collaboration with more nuance?
Either way, it is difficult for me to see a scenario where this does not result in a lot of job losses in the medium term. There are also lots of legal and ethical issues around the data on which the systems have been trained. Those problems aren’t going away, and the leaders of the industry don’t particularly care.
Without leadership or concern from the top, it’s up to us to figure out how to get the algorithms working for us before we’re all working for the algorithms.
The opinions expressed in this column are solely those of the writer, and do not necessarily reflect the views of the editorial staff of ICv2.com.
Rob Salkowitz (@robsalk) is the author of Comic-Con and the Business of Pop Culture.
Source: ICv2
