Skip to content
Home of the finest science fiction and science fact

Editorial

AI Media vs. Human Imagination: Deathmatch?
by Brian Gifford

There are two questions central to the stories and science fact articles that fill Analog’s pages and keep readers coming back every issue: “What if” and “Why not?” Human imagination is what drives our desire to express the worlds that might be or might have been; the things we might have done or may eventually do. “What if?” and “Why not?” are the mantras that move both creative expression and science forward. The only limitations to the fidelity with which we can bring the answers to life are our talent and the tools available to us. But what if the future includes machines that help to bring our imagination to fruition, or why not develop totally new ways creativity might be expressed?

Artificial Intelligence, a fancy name for systems that can learn to find and reproduce patterns from a set of training data, is a tool that’s fast becoming a large—and largely controversial—part of our creative process. AI tools are now adding incredible fidelity and depth to content creation, and just like every technology revolution in history, the effects are shaking up our assumptions, our lives, and the institutions within which we work.

Two types of AI at the bleeding edge of the craft are text-to-image generators and AI chatbots. Image generators like Stable Diffusion and Midjourney can take a text input and produce a unique photo-quality guess as to what the user was asking for. AI chatbots like ChatGPT can parse human language input and produce responses, to include complete products like books and computer programs. AI products often impressively meet the nuance of the user’s intentions, though the current generation struggles to tell fact from hyperbole. A third flavor of AI-produced media is the deepfake, a generated video or audio clip that feels so real that it’s impossible for a person to tell that it isn’t.

Most of us have already heard or seen some of the products of these technologies. Amazing as they are, they could be just incremental improvements. They’re essentially technologies we’ve been seeing for decades, like the CGI that made Jurassic Park and Avatar so impressive. The difference is in the cost and the availability, and in the expertise needed to get a nice product out of the new tools. Free, easy, and no experience needed is a recipe for an explosion of content that even TikTok won’t be able to compete with; or will it? We’ll get back to that in a minute.

First, let’s say we combine these technologies in a Voltron-like exponential increase in capability. We take AI chatbots that can evaluate and respond to nuanced requests from us in an organic way, and we push those prompts to media generation AI capable of creating polished content for us in any flavor we can think of. What if we could ask for, and receive, The Expanse, but with the characters replaced by the cast of Friends? Or The Last of Us, starring Rick Grimes and shot against the backdrop of Vikings? Or three more seasons of Firefly? When I presented this idea to a friend, she believed that the very scope of possibilities would lead to decision overload—a vapor-lock of the imagination. I pointed her to all the content available on streaming platforms and asked her if she had a hard time finding something to watch. Her response was a thoughtful silence.

What will happen to our creative process if we don’t have to fill in all the mundane bits and pieces between our big ideas anymore? What will happen when we can create a world-class finished product just by asking for it? Ask the photography industry.

Since 2003 digital cameras have outsold film cameras, and the industry has never looked back. The changes in technology allowed people to snap away without worrying about the cost of the pictures, fundamentally changing the way photography was approached by both consumers and professionals alike. As the imaging devices got better and smaller, they started becoming an integral part of another exploding technology—the smartphone. Improved quality and simplicity combined with unprecedented accessibility created a tectonic shift in usage that tore down century-old mega corporations that failed to adapt, and simultaneously created enormous business opportunities for those willing to ride the wave of change.

The changes coming with AI are likely to be even wider-reaching than the digital imaging revolution. Computer security and college homework, critical infrastructure design and politics, art and finance—everything is being touched by the changes we’re seeing. Institutional survival will be a matter of flexibility, and wild success a matter of thoughtful prognostication. But let’s zoom in a little and focus on AI media and its impacts.

Streaming platforms and the continued advancement of handheld electronics have revolutionized content delivery over the last decade. These changes have impacted monetization of media, while simultaneously the flow of money has impacted the availability of media to consumers. Consumers pay movie box offices, cable/satellite companies, and streaming providers, with streaming providers steadily gaining market share and traditional TV declining each year since 2017.1 Each of those market segments either produces content on their own or pays production and media companies for content with an eye toward improving their offerings. Wider offerings increase viewership, which improves direct revenue or revenue from secondary sources, such as advertising.

So what changes with AI media entering the scene? First, the ownership of content could become muddied. Does the actor featured in the content make money or have a choice in being included in that content? What if their likeness is the result of an AI’s guess at what they might look like in a given scene? The actor was never actually filmed, but we’ve already crossed that bridge with actors—even the deceased—who’ve been on-screen as computer generated doppelgangers. An actor’s likeness shouldn’t be used without their consent, which should allow them to charge for the privilege. So the subjects of media—or their estates—will have to agree to the creation of media featuring them if the media is going to be legally produced, meaning most will want compensation for being used.

What about the production company that made the media being referenced in the AI’s training data? Their data is being used; shouldn’t they be compensated for their efforts? Will the AI platform need to track the contributing clips in order to figure out who to pay and how much they’re owed? This issue is partially addressed by the way that viewership of content on a streaming platform is tracked. AI content use would just be a more granular look at the same thing.

Finally, what about derivative content creators—those new-model artists that use AI tools to create AI-rendered media for micro-audiences (or not-so-micro audiences)? The lady who renders every season of Doctor Who to star Nathan Fillion as the Doctor, spending the time and effort to track down issues and hiccups to arrive at a clean and quality-checked product? Social media has already solved paying those hardy souls based upon their viewership and subscriptions, making advertising dollars work hard. But can the original content creator sue for infringement or unauthorized use or alteration? With the newness of the technology and its constant upgrading and changing, even the law is racing to catch up.

We’re seeing the fallout of these changes, and the fears that go with them, in the strikes happening right now in Hollywood. Actors and writers, in addition to every other flavor of creative worker, are risking careers and livelihoods to address these changes in a meaningful way, right now. Fears over Fair Use expansion, which would allow the use of derivatives of copyrighted works and individuals’ likenesses without compensation, are driving the entertainment industry into drawing battle lines. The resolution of those fears will have a profound impact on the industry and will inform the direction law and standards of use will take regarding AI use in every other domain as well.

While much of the swirl generated by AI is centered on protecting creative expression, a more tangible drive is money, and nowhere is this more apparent than in advertising. In this strange new reality we’re discussing, advertising space would be everywhere. E.T.’s famous Reeses Pieces? Those could become Skittles, or M&Ms, or Red Bulls. All of the billboards in all of the chase scenes in all of the Fast and Furious movies become useful to whomever is paying the streaming service. Every drink in the hands of the characters of Casablanca become Grey Goose labeled, every vehicle in the Matrix movies changes from GM to Ford. It will be a brave new advertising world.

Even without these new advertising spaces, we’ve already figured out many of the ways to apply dollars to these models. If monetization of the new media has already been (kinda) solved, then all that remains is to provide the platforms, the base-media collections, and the customers. What happens next?

Of course, there will be porn. When I talked through these ideas with friends and colleagues, porn was almost universally the first thing that came up. The adult entertainment industry has been the driver for every major media technology since 8mm film, choosing VHS over BetaMax, DVD over LaserDisc, BluRay over HD DVD, and driving the development of Internet video codecs and APIs.2 The ability to watch anyone do anything will send up flares in the shape of dollar signs to adult entertainment companies the world over, and that multibillion dollar industry is going to get behind AI media and push, hard.

Porn fakes featuring personalities outside of the porn industry will find fertile ground, but that space will be treacherous as well. It’s likely that services will offer personalities the ability to scrape content providers using AI tools to identify their likeness in unlikely places. Requests for content to be removed and warnings about the consequences if a site or service fails to do so will need to become automatic, upgraded versions of the tools used to find social media references to stars today. A revolution in AI-generated lawsuits could follow soon after, but don’t expect that to stop quick-fakes from cropping up regularly.

Aside from porn, expect everything else. And I do mean everything. Favorite show’s been canceled? How about an educated guess at more seasons? How about remaking seasons to fix things you didn’t like? How about adding yourself into the background of every crowd scene, or making yourself the main character, or the bad guy, or the love interest, or the father the protagonist never had? How about updating your favorite movie from yesteryear to today’s standards, or changing the worst movie you’ve ever seen into a rock-musical?

To circle back to TikTok, an enormous amount of money and interest swirls in the chaotic depths that is social media. The relentless pursuit of penny-packet content for the next laugh, the next sexy pose, or the next brain bomb employs millions of influencers whose talents go into producing the dizzying array of vignettes on platforms like Facebook, Instagram, Snapchat, Twitch, X (Twitter), and TikTok. The sheer volume of posts a typical influencer needs to produce in order to stay at the top of their followers’ feeds requires a factory-style approach to creative output. The addition of AI generation to that mix could see an enormous increase in the quality of the media produced by these hardworking creators. So, don’t count traditional social media out of the game as AI rapidly takes over the world.

Another especially interesting possibility emerges when we factor in trends toward diversity and special interest media. Big studios could offer products that can be tailored to the viewer’s ideology, socioeconomic niche, religious preference, or tolerance for graphic and emotional content. Individual users could choose how they want their media delivered to them and alter the delivery in real time to make the art more palatable to them. Imagine a reworked Harry Potter with adjusted plotlines or characters or elements that fans have cried out for in internet chatrooms for decades. Reedited versions of Star Wars and its sequels have already cropped up, but AI technology can make the results more eye-popping and impossible to distinguish from the originals. Hollywood is currently splitting the difference and remaking many successful movies with more diverse casts, leading to a polarizing mix of resounding approval and heartfelt criticism.

The potential variations on media access and scope are limited only by appetite and funding. The results of untutored attempts by individuals may not always, or even usually, be exactly what they were looking for, but I can guarantee the creation of a lot of amazingly entertaining derivative content. However, the impact to the consumer is only the first pebbles of the landslide.

What about jobs? Will the addition of consumer-accessible AI generated media cause a drop in media-industry employment, an increase, or just a change in direction? As seen with media evolutions throughout history, employment is a matter of flexibility. Early silent film stars found themselves struggling to adapt to the addition of recorded sound, causing a serious shift in the who’s who in the film industry almost overnight. The addition of color caused a kerfuffle as well, as did the addition of live broadcast and TV. Video cassettes made home cinema available to the less-than-wealthy and caused movie theaters to pucker with uncertainty, and the change to digital standards like DVD and BluRay at the same time the internet was blowing up pushed piracy to all-time-highs. The shift to streaming video and the multitude of streaming platforms has launched piracy to levels Blockbuster couldn’t have imagined.3 Despite those issues, streaming services are now the driving forces in the entertainment industry, reshaping the entire structure and business model of studios and production companies in just a few years’ time.

The next generation of dominant companies will arise by predicting critical resources and building platforms that take advantage of the new technologies. Being first-to-market doesn’t guarantee success or dominance, but success often goes to early adopters who guess right. Building legally bulletproof access to a wide array of source material for prospective customers would be good start. Nailing down the legal what-ifs and how-can-we for implementing the AI-chatbot supported AI media creation would be another good step.

Individuals looking for opportunities in the new AI technologies should begin exploring prompt generation skills—the technique of providing the input and statements to the AI that get exactly the output the user is looking for. “A monster truck on fire” is a newbie level prompt. “An apocalyptic monster truck blazing with fire flying off the edge of the Grand Canyon into a crowd of zombies” provides a lot more detail. Each AI tool has its own quirks and syntax, and learning how to squeeze the most out of the tools is a skill in and of itself. There are potential careers there for those willing to explore the options and learn the new landscape.

The bottom line is that change is constant, and those who adapt and find opportunity in progress will be successful, while those who attempt to block progress in self-interest or place their heads in the sand hoping that the change won’t come for them may well cease to exist, and almost always cease to matter. Winter, er, change is coming; best to begin preparing.

On that note, it would be good to finish with a brief discussion on the impacts of AI on print media. Recently, author Andy Stanton sold a British publisher a novel he created as a collaboration with ChatGPT.4 The implications of his efforts are sobering, but not unexpected. The piles of submissions that print editors labor through are about to be dwarfed by the avalanche of words that are going to be submitted by AI assisted authors. Initially rules may be put in place that will limit or prohibit AI generated submissions but identifying them will be a Sisyphean task. Besides, if the quality of the writing and the ideas are good, why wouldn’t a publisher want the product, stodgy reactionism aside?

Academia is experiencing some of the same growing pains as AI generated student submissions flood teachers and professors at all levels. The current versions of AI chatbots have some issues distinguishing fact and fiction, and lazy students using the tools to do their homework for them are being caught because their submissions fail the sniff test as a result. However, those issues are rapidly being hammered out, and it won’t be long before the end result will be indistinguishable from a human’s efforts.

One answer in academic circles will be to ensure the questions being answered develop a sort of Turing test aspect to them. Requiring handwritten responses, working on the project in a school-hosted environment, and grading students only on testing performed in class are all stop-gap measures that are being applied now. An alternative is the use of AI to catch AI. If a student’s answers are deconstructed by an AI to reveal the questions that could have been used to prompt an AI response, and those questions are then entered into common AI chatbot tools, the results could be checked against the student’s submission to see if any of the work was taken from an AI tool. Some of these tools are already appearing in the market, and I expect them to be the norm in the near future.

Performing these same checks in the publishing realm may prove to be a waste of time, as the quality and marketability of the content submitted matters far more than the origin of the content. That said, I would expect AI content checkers to be applied in the short term as a way for publishers to winnow their incoming submissions piles as the new tools take off. After that . . . who knows? AI detectors for quality are still firmly in the realm of science fiction for the nonce, though I expect that to change in the indeterminate future.

So, we’re left with almost as many questions about the future of AI in media as when we started. A resounding “prepare to prepare” leaves us with nervous anticipation, but the knowledge that change is coming has become personally relevant. Flexibility, awareness, and a willingness to explore the options will undoubtedly improve everyone’s chances of getting through the revolution unscathed, or even greatly enriched. As a final brain-grenade, what about this article? Would your assessment change if you knew there was AI content in it, or would you still look at it the same way once that seed of doubt was introduced? I couldn’t tell you for sure what my answer would be, but I’d definitely read the article to find out. 

Midjourney prompt: A monster truck on fire
Midjourney prompt: An apocalyptic monster truck blazing with fire flying off the edge of the Grand Canyon into a crowd of zombies

 

Works Cited

  1. Shafer, S., Robson, S., Fletcher, J., Nielson, J., Holden, W., & Leitzinger, P., January 5, 2023, “Traditional media adapts to a digital shift”, https://www.spglobal.com/marketintelligence/en/news-insights/research/streaming-video-revolution-traditional-media-adapts-to-a-digital-shift
  2. Benes, R., May 7, 2017, Business Insider, “How porn has been secretly behind the rise of the internet and other technologies”, https://www.businessinsider.com/porn-behind-internet-technologies-2017-5
  3. Franklin, M., April 19, 2023, Variety, “Piracy Could Result in $113 Billion Loss for Streaming Services by 2027”, https://variety.com/2023/biz/news/parks-associates-piracy-streaming-services-1235587156/

4. Brown, L., April 20, 2023, The Bookseller, “Oneworld signs Stanton’s ChatGPT-inspired adult debut”, https://www.thebookseller.com/rights/oneworld-signs-stantons-chatgpt-inspired-ad

Copyright © 2024 Brian Gifford

Back To Top
0
    0
    Your Cart
    Your cart is emptyReturn to Shop