The Gist
-
AI enhances DAM. Generative AI is transforming digital asset management by automating content creation and improving the organization of assets for faster marketing execution.
-
DAM as a story engine. DAM systems are evolving into story engines, where AI organizes and builds marketing assets based on historical data and brand guidelines.
-
Future of marketing. With AI-powered DAM, marketers can streamline campaigns, creating content from scratch and improve efficiency.
For two years, everyone in the marketing world has been trying to figure out what generative AI can and cannot do. It seemed odd, at first, to outsource work driven by the human need for content, connection and community to an algorithm incapable of experiencing or appreciating any of those things.
However, once marketers saw what AI could generate for pennies, within seconds, the math of ROI overcame whatever resistance remained. How could marketers not load a digital asset management (DAM) system with at least some AI-generated content?
DAM, today, is a system for organizing, finding and distributing content, akin to a library. It will continue to perform that function. We believe it could also become a “story engine” for marketing campaigns. And that could have ramifications for everyone who works in marketing.
How AI Is Reshaping Marketing Jobs and Roles
When ChatGPT debuted in November 2022, its impact on jobs was debatable. Now, it’s clear that generative AI is displacing or changing jobs that, until recently, only skilled creators could perform.
In Harvard Business Review (HBR), researchers recently reported that on freelancing platforms, weekly postings for “automation-prone jobs” fell 21% in the eight months following ChatGPT’s debut. Writing gigs declined by about 30%, while software, app and web development fell by 20%.
In the year following the introduction of AI image generators, like OpenAI’s DALL-E and Midjourney, weekly graphic design and 3D modeling gigs fell by 17%. Over the same period, job postings began to mention “ChatGPT” as a required skill, indicating that “the ability to integrate AI tools into work is becoming increasingly valued.”
AI, in other words, is competing with the people who fill a DAM system with assets. As AI gets more sophisticated, it could graduate from one-off asset creation to getting more involved in campaign orchestration. And we’ve found a good model for how that might work.
Related Article: Navigating the Impact of AI on Digital Asset Management Jobs
The Great American Robo-Novel
To see a potential future for digital asset management, have a look at AIStoryBuilders, a tool for long-form writing. Developed by Michael Washington, AIStoryBuilders builds a database of characters, locations, timelines and plotlines so that ChatGPT can help with writing. Essentially, it learns from an initial set of human-generated assets to create new ones.
On the Hanselminutes podcast, Washington describes the limits and potential of his software.
“I found out that it's only good for writing one paragraph at a time, and I still have to keep editing that paragraph,” says Washington. That paragraph doesn’t come easy either, he adds. “I have to build those characters up. I have to say the locations. I have to put the timelines in. There is a lot of building up.”
He's describing the work that goes into “grounding” an AI into the reality of a story, so that it doesn’t hallucinate. Once this information is organized in a database, the AI can learn and relearn this info through retrieval augmented generation, or RAG, before spitting out a paragraph. That way, it gets basic facts right, like the color of a character’s hair, the setting, the year and so on. ChatGPT can even fine-tune a model to emulate someone’s writing style. It only needs about three chapters of writing to do that, Washington says.
Is generative going to write the next great American novel? Not on its own. “[AI] has no soul, it has no desires, it has no creativity,” Washington says. “But if you organize things in a certain way and feed the AI through the grounding, it actually can be helpful.”
How Digital Asset Management Powers AI-Driven Campaigns
A marketing campaign is a story of sorts. It’s about a product (what) created for an audience (who) that needs or wants it in a certain context (when/where), and the product’s function (how) helps to deliver a value proposition (why).
If AIStoryBuilders can organize characters, locations and timelines into a database, then surely an AI-powered DAM could organize the elements of a marketing campaign and build upon them. The database would include product information, personas, historical assets and perhaps performance data from past campaigns. These could be communicated to AI through RAG and used to fine-tune a model that adheres to the brand’s voice and guidelines.
Initially, the DAM story engine would perform best on finite tasks, same as AIStoryBuilders. Make a 15-second YouTube ad, an Instagram post or a TikTok video. Develop an email template. Write the first email in a drip campaign for people who shared their email address for a chance to win a free product. It could spit out mailers, TV ads and radio ads, too.
Could generative AI run a marketing campaign from start to finish? Maybe one day. The newest version of Anthropic’s Claude can “use computers the way people do — by looking at a screen, moving a cursor, clicking buttons and typing text.” Surely, it could operate ad platforms and marketing automation platforms.
Plus, with access to AI influencers and AI models, the DAM story engine could generate campaign assets without waiting on human beings to photograph models or ship products to TikTok stars. Like a driver in the seat of Tesla on autopilot, the human marketer would need to be vigilant — and ready to take the wheel, if necessary.
Related Article: 24 Enterprise Digital Asset Management Solutions Examined
Overcoming Obstacles to AI Integration in DAM Systems
Digital asset management is an ideal system for an AI story engine because it contains content and metadata — rich information about that content and how marketers expect it to be used.
Still, there are some significant obstacles to overcome:
- First is the cost. Generative AI companies are incinerating cash and selling tokens at or below cost. Unless generative AI companies develop more resource-efficient models (or cheaper chips and electricity sources), there is a good chance costs will rise to the point where a DAM story engine isn’t economical.
- A second obstacle is the pace of progression in generative AI. Until recently, generative AI reliably improved with scale; give a model more data and processing power, and you’ll get more intelligence out of it. Now, though, OpenAI employees admit that progress has slowed.
- Third is the concern about image quality. DALL-E, for instance, still produces airbrushed, cartoonish imagery and struggles to render text in images. While the output is good enough for a Slack joke or holiday card, it’s a far cry from what professional designers do. Then again, OpenAI’s Sora model for video generation seems to produce lifelike imagery, so maybe DALL-E has hope.
Regardless of these barriers, we expect more of the creative workflow to take place in digital asset management systems, which increasingly will become a database for organizing stories and an engine for creating content. We also expect marketers and creatives to become skilled generative AI users.
Generative AI may not understand and experience content, connection and community like human beings, but we do. And maybe that is the most important qualification we’ll have in the era of DAM story engines.
Learn how you can join our contributor community.