Article

Ensuring scalable and accountable AI in media

July 09, 2024 – 7 min. read

Successfully establishing accountable, transparent, and scalable use of AI in a media enterprise is not an easy achievement. The heavy lifting of having to deal with disconnected, siloed AI services often copied/pasted out of a browser and into the story creation interface is becoming a reality, as newsroom staff discover and start using new AI tools.

Mark Van de Kamp has seen his share of innovations in the media industry, being on the technology side of media for almost 21 years, currently as the Chief Innovation Officer at Stibo DX, the company behind the media enterprise platform, CUE. The company recently launched its new AI platform, called CUE Autopilot. It’s architected as an abstraction layer on top of the CUE platform, making AI implementations scalable and easy to govern, across brands and titles.

We sat him down for a conversation on the scalability of AI.

Mark, AI has sparked various reactions across our industry, with sentiments ranging from admiration to skepticism. Could you sum up your stance on AI’s role in news media today?

Mark: Yes, absolutely. When OpenAI introduced its large language model in November 2022, it quickly became clear that this phenomenon would be quite disruptive for our industry. At Stibo DX, we sensed both an appetite and a need to share ideas on how to cope with it in our CUE community. There is a fear of missing out on opportunities, as well as a fear associated with the risk of deploying a brand-new category of technology in the newsroom, and I get that. AI is moving very fast at the moment, yet at the same time, I believe we are at a turning point where AI now has to prove its value to the common journalist and not just to superusers.

At Stibo DX, you work on making AI scalable for media groups with a consolidated media platform. Can you share a few words about your approach to this?

Mark: For us, as a vendor of software solutions, it quickly became clear that AI will impact how we develop the platform. So, we had a strong need to learn and distinguish between what is possible and what is valuable. Given that our CUE community mostly consists of broadcasters, media groups, and media conglomerates, the conversation quickly turned to the question of what AI at scale looks like.

It was natural to involve our community even more in charting the course for product innovations. For example, our print automation solution is built with customer development partners, enhancing our capability to define and develop the solution, both from a functionality and UI point of view to a whole new level. However, for AI, we wanted to do something different and decided to create a customer AI advisory board. As a stable, long-term partner for our customers, we cannot allow ourselves to be dazzled by the potential we see in AI, and the advisory board kept us grounded throughout the development of CUE Autopilot.

How has your advisory board influenced the understanding of using AI in the newsroom?

Mark: Our advisory board has played a large role in our understanding of the potential of AI within newsrooms. A crucial insight from our discussions is the value of simplicity in AI tools. Rather than opting for the most comprehensive and sophisticated tools, we focus on selecting and implementing workflows for using high-quality, easy-to-use tools. These tools are designed to minimize repetitive tasks, which is essential for efficient workflows. This approach enables users to create content more easily and efficiently in a controlled, reliable, and well-governed environment.

In our broader discussions on the board, while we explore deeper topics, our primary emphasis remains on ease of use, predictability, scalability, and the seamless integration of AI into existing workflows.

Why is governance of AI so crucial in large media organizations?

Mark: Effective governance including traceability is essential for the responsible use of AI tools – it aligns both legal standards and ethical norms, ensuring that AI-driven initiatives are transparent and uphold the trust the media is given by its audience. 

Moreover, governance ensures the scalability of AI tools. This means that these tools can grow in capacity and functionality without compromising their performance, supporting the organization's broader objectives. It applies a strategic role to governance as it also facilitates the coordination of AI development across various departments. As a result, it not only facilitates an upgrade of workflow efficiency in the singular newsroom, but an upgrade of efficiency across the entire value chain in the group.

How does CUE comply with the need for governance of AI?

Mark: We recently launched the new AI product called CUE Autopilot, which is a result of the input and feedback we received from the AI advisory board sessions over the past year.

CUE Autopilot ensures that the use of generative AI in all phases of the content workflow remains managed and traceable and that no AI-generated content goes public without human approval. When new AI-powered tools emerge, integrating these into the CUE workflow is seamless, and you stay in control of your journalistic integrity and regulatory compliance. On a practical level, it’s an abstraction layer on top of our CUE media enterprise platform that deals with AI implementation across the three products that are part of our platform: CUE Content Store (CMS), CUE Print (Print CMS), and CUE DAM (Digital Asset Management). With a consolidated platform like CUE with an AI layer on top, we secure the ability to scale, govern, and access AI on a group level as users and titles are onboarded to the CUE solution decided on the group level.

What’s next in AI, and what are the main developments on the AI roadmap for CUE?

Mark: It’s important to emphasize the difference between new AI tools provided by various providers and the infrastructure to support the use of these inside the CUE platform—our responsibility is focused solely on providing the well-governed integration infrastructure when new (valuable) AI opportunities arise. So, while we may be excited about the upcoming possibilities in automated video generation, text-to-speech, etc., it’s their use in CUE that we are investigating and developing.

One area that we are soon ready to expand on is the improved reporting and tracking of how AI is used in the newsroom, e.g., how many articles have been touched by AI, which paragraphs, fact boxes, etc., were generated, and other similar data points. Management needs to be able to monitor this.

Another important topic is ensuring that our AI solutions enhance the journalistic process rather than complicate, the journalistic process. Good UI is a moving target, and we do a lot of testing and iterating to make sure that the user experience for journalists working with AI remains clutter-free and efficient in CUE.

Read more about CUE Autopilot: https://www.stibodx.com/products/cue-autopilot

Read the guidelines Stibo DX uses when working with AI: https://www.stibodx.com/resources/charting-a-responsible-course-in-ai-for-news-media