Editorial
The Gist
- Job evolution. AI's rise challenges traditional roles in digital asset management, requiring new skills and adaptability.
- Admin essentials. Despite AI advancements, DAM admins remain crucial for system order and efficient AI integration.
- Training imperative. Effective AI in DAM needs continuous training and human oversight to ensure accuracy and relevance.
Editor's note: This is the first of a multipart series exploring the impact of AI in DAM, content management and across various martech fields and beyond.
In the 1990s, the software industry began releasing products so integral to business that they required their own administrators. People built careers around systems for enterprise resource planning (ERP), customer relationship management (CRM), project management and digital asset management (DAM). The rise of digital asset management jobs provided new career opportunities.
And it took only about 30 years for AI in DAM to throw those careers into uncertainty.
The Impact of AI in DAM
The question facing everyone in software administration jobs — and office work generally — is what our jobs will consist of as AI becomes increasingly proficient at our tasks. DAM administrators are canaries in the coal mine.
Years before ChatGPT debuted, DAM vendors began integrating image recognition AIs to semi-automate metadata tagging. DAM admins remained indispensable despite that innovation — and I believe they will remain indispensable even as large language models (LLMs) automate more of the DAM workflow and affect digital asset management jobs.
Today, a DAM system without a DAM administrator is like a library run by its patrons: chaotic. AI alone cannot bring order to a DAM system, but DAM admins without AI may find themselves struggling to preserve order.
Let’s take a look at the impact of AI in DAM and explore why.
AI as Pattern Finders
No one knows why LLMs can do the miraculous things they do — seriously. But we do know that AI excels at finding and predicting patterns based on examples we feed them. This pattern-finding skill is why AI image recognition found early adopters among DAM admins. It showed potential to relieve admins from hours of reviewing and tagging raw assets, though in practice its vocabulary and descriptive powers were limited.
But if AI learns the wrong pattern, it can become highly efficient as mislabeling assets and mucking up DAM systems. AI needs DAM admins more than DAM admins need AI.
Earlier this year, I partnered with Acquia integration developer Jacob Williamson to study whether a standard GPT-4 model could not only tag keywords (like image recognition AI) but also draft basic product descriptions. We engineered six prompts designed to elicit descriptions and keywords. We then collected product images from six categories and used a script to run the images by GPT-4. We evaluated accuracy — whether GPT-4’s description identified the contents of the image correctly — and precision — whether the 10 keywords generated for each image were factual, relevant and plausible as DAM search terms.
The highest average accuracy score for any prompt was 88.3%, meaning it misidentified more than one in 10 product images, while the highest average precision score was 92.2%. Had we used the best prompt on, say, 20,000 images, 2,340 of the descriptions would have been inaccurate, and 15,600 keywords would be imprecise or even misleading.
Thus, our study (coming soon in a peer-reviewed journal) highlights that LLMs need oversight and training before they modify assets at scale. Who better to do that than a DAM admin?
Related Article: Examining 19 Enterprise Digital Asset Management Solutions
The Tag Evolution
Our study testing GPT-4’s DAM administration skills suggests that with training, GPT-4 and its ilk could become proficient at keyword tagging, product descriptions, alt-text and more. The challenge is that AI content creation is disrupting existing metadata schema — and AI in DAM itself cannot rewrite that schema for a business environment in which cultural and legal expectations for AI change quickly. That still requires a true DAM admin.
To illustrate, the Wall Street Journal recently reported that models are selling their likeness to platforms like Fashion AI to reduce the stress and instability of their in-person work. Models retain control over which brands can use their likeness and at what price. AI can then render the model in different clothes, poses and places. With the same tech, brands could capture AI versions of their online influencers, employees and happy customers to create content faster and at lower cost.
However, DAMs aren’t yet set up to manage these hybrid AI models. How many images are you allowed to create with their likeness, and what counts against that limit? For example, reusing an image with tiny modifications to the background, rather than changes to the models themselves, may or may not count. Or for how many seconds can the model appear in a video before a brand incurs additional fees, and how is that tracked and enforced? Could someone’s likeness be altered to the point where it is no longer “them”? Could brands add AI voiceover to an AI model’s video, even if it’s not that model’s own voice?
As AI in DAM changes jobs, DAM admins must be at the forefront of strategizing how to tag and track assets for these strange and novel scenarios. AI is not equipped to manage these issues.
Related Article: Why Customer Experience Pros Should Care About DAM
AI Begets AI
In my view, one thing that will change concerning digital asset management jobs is DAM admins will feel pressure to use AI if they don’t already. Generative AI brings the marginal cost of content creation to near zero. In other words, making five images with AI is barely more expensive than one making one image with it — because each image takes seconds, and typical LLM subscriptions charge monthly, not per token. DAM admins will soon see a bombardment of assets that, as discussed, have complex metadata and image rights considerations.
That means the central role of a DAM admin when it comes to AI in DAM will be to train and tune AIs such that they actually help and keep up with this inflow of content. Well-established practices for feeding AI examples and giving feedback on its output will become part of the DAM admin skillset.
Let’s not forget that for businesses, the point of a DAM system is to reduce time-to-market, increase content ROI, ensure brand consistency, and increase “content effectiveness” (which brands can define quite differently). Speed and consistency are the underlying themes.
As layoffs in big tech already hint, employers are curious to see what employees can do if, instead of hiring more people, they have existing employees harness AI. Simply put, DAM admins must become skilled AI managers as AI in DAM evolves.
Webinar
Sep
25
CMS Briefing: A Live Look at What’s Next in AI-Driven Platforms
Learn how leading organizations are using AI‑driven tools to publish faster, personalize smarter and stay secure.
Register
Webinar
On demand
Ready or Not: How Data-First Organizations Are Unlocking Agentforce Potential
Learn how to cut through the noise, activate Agentforce and build a Salesforce AI strategy that actually delivers.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
Webinar
On demand
From Hype to High-Impact CX Strategies That Actually Scale
Turn buzzworthy AI and outsourcing trends into measurable CX wins with fresh 2025 data.
Watch Now
Webinar
On demand
Insights to Action Rethinking the Contact Center for Real Business Impact
Join our exclusive webinar to hear CX executives share their innovative strategies for transforming service delivery.
Watch Now
Webinar
On demand
Accelerating Healthcare Ops with AI
Sign up to see how your peers are putting AI to work — and making it stick.
Watch Now
Related Article: Why Personalized Experiences Need a DAM at the Core
What's Next for AI in DAM
Those of us in office work will be in a similar position to DAM admins who are adjusting to AI in DAM, sooner or later. AI will become proficient at tasks that once consumed hours of our days. Even if the AI isn’t as good at the task as we are, the gain in efficiency might offset the loss in quality. We will train and use AI to optimize efficiency, and we will exercise human judgment to keep AI from failing, or scaling failures.
Hopefully, a career managing AI will be engaging. The DAM administrator is evolving in that direction, and so is every job at a computer screen.
Learn how you can join our contributor community.
