Experiences, Insights & Challenges Regarding the Fusion of AI and DAM

This feature article has been contributed by IT consultant, project manager and DAM specialist Martin Reinheimer.


AI-based solutions are the IT trend of the past year. The biggest gamechanger for this has certainly been new, simple front-ends, which now give every end user the opportunity to try out the various tools and gain their own experience. And the solutions are numerous, whether they are generative AI tools such as ChatGPT for text generation or, for example, Midjourney and Firefly for image generation, or assistance systems such as the Copilot offered by Microsoft.

As is the case with every innovation, there is often a long start-up phase in which such tools are evaluated in research and development before the hype begins, shortly before they are ready for the market and interest rapidly grows.  However, this is often followed just as quickly by a slowdown, especially if the innovative idea does not bring any economically significant added value.  This is certainly not the case with AI, and although the initial euphoria is slowly subsiding, it is clear to see how much the trend is spreading – not just in DAM or MarTech –   and finding its way into companies.

But let’s take a step back – what are we talking about? When we talk about DAM, what does AI actually mean?

AI aims to simulate human-like intelligence.  On one hand, this is based on machine learning, in which models are trained using structured sample data, and on the other using deep learning, in which unstructured data is processed in neural networks.  In both cases, the result is that the trained model contains knowledge that enables it to automatically recognise image content or make predictions, for example.

AI is not a new thing.  The Dartmouth Conference, in 1956, when the term was coined for the first time, is considered the birth of AI.  However, the breakthrough only came after the turn of the millennium, when machine learning picked up speed thanks to ever better hardware and, in particular, in the last 10 years, when big data and scalable, cloud-based computer architectures made it possible to use deep learning to adapt so-called large language models (LLMs).

So, what aspects of AI gives users of DAM systems and manufacturers a market advantage?

In terms of DAM solutions, there are several use cases where the benefits of AI are obvious:

  • The better the AI is trained and can therefore automatically tag and classify assets, the more obsolete the time-consuming manual tagging of assets becomes – this is particularly exciting when it is not only applied to static images, but also to video content, which can be classified and segmented fully automatically.
  • The uniqueness of the assets is guaranteed when AI finds and merges duplicates.
  • Finding the right assets quickly becomes easier the better AI understands the search query and can perform natural language processing.
  • By recognising a point of interest in images, dynamic cropping for the respective output channel is possible without losing important image content.
  • By combining several AI tools, new workflows are possible, for example to add keywords to images, which are then converted into a promotional descriptive text about the image content, and then potentially translated into any language.

AI does not end with DAM in the MarTech infrastructure.  Further opportunities open up, especially in other marketing processes, where the entire customer journey is taken into account:

  • Automation: AI makes it possible to increase the degree of process automation by updating and distributing content automatically.
  • Individualisation: Based on available information, including the purchasing behaviour, customers can be provided with tailored, individual advertising offers.
  • Predictions: AI-based analyses can be used to derive trends and align content production with these trends.

But how do you start using AI to improve your own market position?

In order to realise the competitive advantage of AI, you need to know your own situation and be aware of the state of innovation.  It is no different to the traditional continuous improvement process, which should be practised in every part of an organisation.

This involves regularly analysing your own production processes and identifying and evaluating potential for improvement.  External advice can help to take an objective look at things; for product-specific innovations, the software manufacturers should be consulted to find out how the developments of new releases can be utilised profitably.

Despite all the euphoria surrounding the topic, there are of course areas where there is still room for improvement.

Yes, it is incredible how generative AI is able to magically create content out of nothing.  But as the importance of the details in the generated content increases, so does the need to specifically train the underlying model.  If you want to generate a clip to promote your own products, for example, you quickly realise that logos or other important brand recognition features will not be displayed correctly until the model has been sufficiently trained with regard to these specifics.

When it comes to pattern recognition, you must accept that despite extensive training, the generation may still not be able to provide the same pattern recognition performance as a human.  It is to be hoped that this situation is only a passing phase that will disappear with the rapid pace of development.

Nevertheless, if you look back, the new tools are incredibly powerful: if you have the task of enhancing the contents of an image today, this can be done fully automatically by AI.  Just a few years ago, it would have taken hours to painstakingly modify the image in a desktop publishing application.  And if we look back even further, when producing on negative film, it was only possible for specialist artists to accomplish such a task.

The legal component

One dilemma with innovations is always the legal component: Is the use of such new tools even permitted?  Are manufacturers allowed to incorporate certain functions into their products?  The technical possibilities are often one or two steps ahead of the legislation, which does not always ensure safe and responsible operation.

Here, too, it is important to be courageous in order to secure your market position, but also to keep an eye on the legal framework.  The European Union presented a draft of the AI Act back in April 2021.  Now, three years later, the regulation is set to come into force.  However, it is uncertain whether such regulation will be implemented at a global level.

There are situations where you often have to get an overview of the grey area first: what about copyright issues relating to AI-generated content?  The Content Authenticity Initiative (CAI), for example, is working with its members to create a standard that provides reliable information about the origin of digital assets.  Is facial recognition in image and video files a function in my DAM system that I can use in accordance with GDPR?

Currently, there is no clear answer to these questions.  You always have to see it in context, i. e. to what extent does it affect the content I use, and which laws and regulations apply to me?  As a company, it is important to define a framework or usage policy for employees in a kind of codex.  This requires that the company has already dealt with the innovations and continues to do so.  Where does the processing of my own content by an AI tool take place?  Would it be an option to train a specific AI tool in-house and host it on-premise to ensure that my internal company data cannot be seen and used by the outside world?

The key to all these questions is to stay on the ball and adapt your own policies in line with innovation and the current legal situation.

And will employees become superfluous in the new AI world?

Whether you’re a photographer or graphic designer, AI will of course take over repetitive work and can also act as a creative stimulus in the field of generative AI.  Nevertheless, in the foreseeable future, no-one will lose their job if they start to explore the opportunities that are available.  Every creative person must continue to develop and find ways to utilise the new tools for themselves.  You can then regard AI tools as additional tools in your own workspace that help you to complete your tasks even better.

However, there are also areas where the work will change more: Automatic translations are currently so good that full automation can be used for everyday tasks, for example for image description texts in a DAM.  Translators will therefore have to specialise further in order to work in areas that require particular specialist terminology or legal certainty.  Alternatively, it is possible to take the bull by the horns and look more closely at the possibilities of improving your own productivity with the use of AI tools.

It is interesting to note that experiments have shown that the creative tasks in particular are not completed faster through the use of AI, but rather that AI is used as another tool in the toolbox.  This means that AI does not necessarily reduce the price of creative services.  Nevertheless, it is significant to note that consumer studies have shown that the willingness to pay decreases when the customer is informed about the use of AI.  It remains to be seen whether this behaviour will also regulate itself at some point through the increased use of the tools.  Until then, it remains an exciting ride on the waves of AI!


About Martin Reinheimer

Martin Reinheimer is the Founder of proCress – a recently established company specialising in IT project management, consultancy and martech (marketing technology).  Martin has been working at the interface between IT, marketing and media production his whole career

You can connect with Martin via his LinkedIn profile.  You can also view Martin’s profile in our DAM expert directory.

Share this Article: