As Adobe Joins Generative AI World, Hollywood Grapples With New World

It was fitting that Adobe
ADBE
chose Las Vegas this past week to announce Firefly, its entry in the red-hot generative artificial-intelligence race. As one corporate consultant working the exhibition hall put it, such generative AI functionality is now “table stakes” for every big tech firm, a minimum (if hefty) bet that is quickly becoming a must-have for both customers and investors.

Firefly is very much in beta still, as Adobe executives repeatedly emphasized. Would-be users must sign up online, and will be given access over the coming weeks to tools that make it easy to swap in image backgrounds, products, props, colors, marketing text and more, then quickly create and publish multiple variants for other platforms.

Firefly’s launch represents another big, if still preliminary step, putting Adobe products alongside high-profile offerings such as ChatGPT, Midjourney, and Stable Diffusion from companies such as Microsoft, Alphabet and OpenAI.

Generative AI is breaking out in lots of places of a sudden. Last week’s episode of comedian and game-show host Drew Carey’s SiriusXM radio show, Friday Night Freak-Out, was recorded using an AI voice created by Eleven Labs and ChatGPT, according to a release from the channel where the show runs, Little Steven’s Underground Garage.

And other examples are multiplying quickly, but Adobe executives repeatedly said caution as much as excitement is warranted.

“Machine learning and AI are powerful but also require a really thoughtful approach that can amplify and not replace human creativity,” said Adobe CEO and Chairman Shantanu Narayen during the Adobe Summit. He called content creation “a massive opportunity for growth for every company.”

Adobe made several other AI-related product announcements, all designed to more tightly integrate pieces of the company’s traditional creative tools such as Photoshop and Premiere Pro with its newer marketing and data-analysis tools.

“Adobe 100% stands behind the belief that this technology is an accelerator for creative uses, not a replacement,” said Ely Greenfield, Adobe’s CTO of digital media, in an interview. “These technologies become another tool that can accelerate the production process, but don’t replace it.”

The Adobe announcements were well received among the companies and consultants at the conference who use its tools with thousands of clients trying to sell pretty much every imaginable product or service out there.

“It’s a new way of looking at marketing,” said Phil Regnault, who leads PWC’s
PWC
Adobe practice in advising companies how to use its many tools. “It used to be just PDFs and Photoshop. Now they have a number of applications in the (Chief Marketing Officer) suite. (Customers) need to have digital literacy, then codify that in house. I think it may just be responsive to what their customers are looking for.”

After half a decade focused on collecting, analyzing and acting on data for corporate suites, the Adobe moves are designed to fix another set of issues, what Regnault called a “log jam” on the content creation side. Marketers can’t make enough content to keep up with all the platforms and market segments now reachable.

The Adobe announcements also mark an important step for one of the companies that professional creators in Hollywood and elsewhere rely on to make movies, TV shows, commercials, visual effects, podcasts, online publishing, and the like. Premiere Pro, Photoshop and other Adobe products already had modest AI-powered capabilities, to ease repetitive tasks such as transcribing video or suggesting the best look-up table settings for a photo.

Greenfield emphasized the limits of AI, even amid its power to free human workers from a lot of grinding repetitive work needed to churn out material needed for cutting edge data-driven marketing.

“These technologies so far, they’re not great at non-linear thinking, but they can spark non-linear thinking,” Greenfield said. “They can help people explore. They can reveal things that were inspired by other content.”

That’s exactly the concern across Hollywood, where actors, directors, and, especially, writers, are wondering where they’ll fit in this AI-fueled content-creation future. It was appropriate that Adobe invited Oscar- and Emmy-winning writer/director Aaron Sorkin (The Social Network, The West Wing) to speak at the conference, though he opened his keynote conversation asking, “Why am I here?”

As Sorkin pointed out, “Somehow my job changed over the last few years, and I’m a content creator now.”

The use of generative AI tools such as ChatGPT have stirred Hollywood creatives in a big way. The Writers Guild of America just started negotiations with the major studios for a new contract (the current one ends May 1), with a “pattern of demands” trying to define how and whether studios might use AI tools to create or rewrite scripts.

Some companies are already touting their AI tools trained on thousands of existing film and TV series scripts (copyright lawsuits have already erupted), but Sorkin said that’s likely not to be as successful an approach as backers might suggest.

“Knowing what people want and giving it to them is a bad recipe for storytelling,” Sorkin said. “There are hundreds of ways to prepare beef. But if I prepared beef in the way most people consume it, it would be a McDonald’s
MCD
hamburger.”

Storytellers are “leaders, we’re not followers,” Sorkin said. “Writing is hard, at least it is for me. I try to write what I like, what I think my friends will like, then I keep my fingers crossed that enough other people will like it that I’ll get to keep writing.”

The Writers Guild has plenty of other contract demands this time around, many of them left over from 2020’s pandemic-truncated round of negotiations.

The streaming services, with their reliance on shorter six- to eight-episode seasons, and fewer seasons overall, have been one big source of complaints. It’s hard to make a living on a per-episode basis when seasons are a third to a fourth as long as old-school TV, especially when residuals and syndication are also sharply limited.

There are other issues, like contract holds that make it hard for writers to get a second job, and the growing use of “mini” writer’s rooms that mean fewer staff positions overall.

The possibility that penurious studios would use AI tools to fluff up a script, or write it in the first place, is a new issue but a big one for writers who are already worried about disappearing jobs.

Sorkin, who wrote movies about tech titans Steve Jobs and Mark Zuckerberg, said his own writing was transformed by a different technology revolution, when he cobbled together enough to buy an early Mac computer in the late 1980s. Copy and paste, and a delete key, were magic for a struggling young playwright working a “survival job” as a bartender in a Broadway theater. More recently, he also directed the last three movies he wrote, something made possible by, yes, lots of tech.

“Technology that’s been developed has allowed people like me to make movies I wouldn’t have been allowed to make 20 years ago,” Sorkin said. “I think technology can be a co-pilot too. I have to sit in a room with people who are experts at it (while I don’t) know anything.”

But there are limits.

“Where I get nervous about technology, where I don’t think it is as useful is when I hear about software that can write a screenplay for you,” Sorkin said. “A computer didn’t write those screenplays that are being fed into the machine in the first place. I think you’re going to enjoy things written by humans for a long time.”

For other people, the new technologies promise new possibilities, even new careers.

“If you want to sound more Sorkin-like, these technologies can help,” Adobe’s Greenfield said. “Aaron Sorkin may have some things to say about that.”

Indeed, if not Sorkin, it’s not hard to imagine another name-brand talent of his caliber who might license the markers of his writing (is it possible that Chris Rock will have yet more ways to be omnipresent?). That could create a passive income stream by training up an AI to make someone’s next TV commercial or industrial video a little more “walk and talk” with idealistic, aspirational language and ideas.

Much more likely, Greenfield said, is “a growing business of prompting,” the people who have learned the intricacies of one or more specific AIs, and know how to coax the best text, imagery or video from a specific, generally iterative set of “prompts,” or instructions to the AI.

“I think we’re still a reasonably long ways away from a lot of mid- to high-value use cases that can deliver well without human involvement,” Greenfield said.” If I’m not there to review (the outputs from the AI), that sounds really scary. These are assistive tools to humans, and I think it will be for a while.”

The new business is in a place much like search-engine optimization, which didn’t exist as an industry 20 years ago. Now it’s a multi-billion-dollar sector.

“It might be a moment of arbitrage,” Greenfield said. “We’re driving this deeper into traditional workflows, and into new workflows. It’s less about using arcane language, and (more about) being able to drive multi-modal outputs” across a variety of distribution platforms.