AI and DAI: Pursuing greater dynamism in machine learning and advertising

Broadcasters are closely watching recent advancements in AI and machine learning (ML) technologies and their use in areas such as Dynamic Ad Insertion (DAI). These advanced technologies have the potential to reshape the way advertising is delivered and consumed, driving increased relevance, efficiency and revenue for both broadcasters and advertisers.

Current Use of AI/ML in Dynamic Ad Insertion

AI and ML are already impacting DAI by driving real-time decision-making and sophisticated targeted advertising. Daniel Pike, Chief Product Officer at Covatic, said: “AI/ML algorithms analyze vast amounts of data, including user behavior, preferences and context like time of day and content consumption patterns, to dynamically select the most relevant ads for each impression. These technologies can predict which ads will perform best based on historical engagement metrics, making ads more relevant and effective.”

AI and ML have the potential to go beyond basic demographic targeting by leveraging numerous data points to build a comprehensive understanding of individual users. This can include factors like browsing history, social media interactions, and even real-time geographic location. “Such models also come with several practical, legal and ethical challenges,” Pike warns. “One of the challenges is user privacy,” he says. “Decentralized AI and ML have a significant privacy advantage over traditional approaches because personal data can be kept with individual users and doesn't need to be pooled in a big data lake somewhere.”

read more How can streamers make the most of DAI and avoid the complexities of GDPR?

Johan Bolin, chief business officer at Agile Content, points out that current systems already use some AI, but that it could become even more sophisticated. “The definition of AI is very broad,” he says. “Machines that match user profiles with relevant ads are established; next is tracking ad efficiency and feeding that back to ad decision servers to 'learn' which ads work best in what situations. Some of this is 'private magic' within various ad insertion systems, but I think Google, Amazon and Facebook have already achieved this to varying degrees.”

Effectiveness compared to conventional algorithms

There are many effective algorithm-based ad targeting solutions available today. The question is whether AI/ML-based targeting can offer a notable advantage over current approaches. Pike explains: “AI/ML-based targeting is more effective than traditional algorithms because it can process and interpret vast datasets quickly and accurately. Traditional methods often rely on static, rule-based systems that can miss nuanced user behaviors and preferences. In contrast, AI/ML continuously learns and evolves,” he says.

Bolin agrees: “How effective it is will depend on how it's applied, but we can assume that with enough data and enough advertising, this could be very effective,” he says. “Given that advertising is the core business model for some of the biggest tech companies, it's reasonable to assume that a large portion of the huge budgets allocated to AI are actually aimed at increasing revenue through better advertising.”

Addressing the black box problem

One of the significant challenges of AI/ML in DAI is the so-called “black box” problem: often, even the developers of the AI ​​technology don't understand how their systems achieve certain results. This lack of transparency can make it difficult for advertisers to understand why certain ads are chosen over others, or to verify that the AI ​​system is working correctly.

Pike emphasized the importance of transparency and accountability in AI systems, saying: “AI solutions can be difficult to fully understand, conceptualize, and explain.” “There’s always the risk that the AI ​​will fill in the gaps in its capabilities and provide answers that don’t make sense – that is, a smart AI will provide plausible but false results. If you don’t know how the AI ​​arrived at its decisions, how can you tell a good AI or ML solution from a bad one? Looking at the inputs to the system is a reasonable starting point. For example, if you’re trying to understand a user’s age, gender, and purchasing preferences, but looking at the input data, you can’t see any way that a knowledgeable person who has invested enough time and energy into making a sensible decision based on that data, it’s probably fair to say the AI ​​can’t do it either. If you have a set of truths and you can, thoroughly test the performance of your system. Test, test, and test some more.”

Computing and Energy Impact

Powerful AI/ML capabilities come with significant compute and energy demands. Training and running AI models, especially deep learning algorithms, requires significant compute resources. Traditional algorithms are less complex and therefore consume less energy.

Bolin points out that determining the resource footprint of such a system is like asking, “How long is your rope?” and that it all depends on what you want the AI ​​to do. “In a pretty basic form, you just match user profiles with the right ads,” he says. “But if you imagine in the future that the entire ad is created 'on-demand,' targeting users with generative AI, and the ads are encoded on-the-fly into videos as product placements, then you end up with a lot more energy consumed not only in the AI ​​creating the ads, but also in encoding all the streams and ads.”

The challenge for the industry is to balance the efficiency gains enabled by AI/ML with the need for sustainable energy consumption. Edge computing, which processes data closer to where it is generated rather than in centralized data centers, offers a promising solution.

Pike explains: “Edge computing and specialized AI chips can help reduce the computing and energy demands of AI/ML processes by optimizing performance and energy efficiency. Distributed AI is definitely an exciting new frontier.”

read more DAI and Sustainability: Session-Based Solutions and Ad Delivery Efficiency

Revenue potential for broadcasters and advertisers

Paik believes that AI/ML-based targeting has the potential to significantly improve revenue for broadcasters and advertisers. These technologies can increase the value of ad inventory by serving more relevant and engaging ads. “AI/ML-based targeting has the potential to significantly improve revenue for broadcasters and advertisers,” Paik says. By serving more relevant and engaging ads, AI/ML can increase engagement and conversion rates. This precision targeting means that ad inventory is used more efficiently, potentially increasing the value of each impression. While initial gains may be incremental as the system learns and optimizes, the long-term benefits will be substantial.”

Bolin also believes that the use of AI has great potential to increase revenue by changing the very nature of how products are promoted. “Especially with more advanced and long-term implementations of Gen-AI,” he says. He envisions the technology being used for ad creation and product placement. “Perhaps by extending some scenes with ad scenes or sequences rather than traditional interruptions,” he says. “This will be a way to incorporate much more advertising into content without increasing the intrusion of the content. This approach increases the size and value of the inventory while improving the experience, creating value for all parties involved.”

Trust and Transparency

Generative could potentially create real-time ads specific to specific audiences by leveraging data about user preferences and behavior to make ads more relevant and engaging.

But this also raises concerns about intrusiveness and user trust. Paik stresses that a careful balance is needed between personalization and privacy, ensuring transparency and user consent to avoid negative perceptions. “Generative AI holds promise for creating highly personalized ad content, but it walks a delicate balance between relevance and intrusiveness,” he says. “On the one hand, generative AI can create unique, personalized ad experiences in real time, which can significantly increase user engagement and effectiveness. On the other hand, if not handled with care and transparency, such personalized ads can be perceived as creepy or intrusive, undermining user trust. It is essential that the industry balances personalization with user privacy and consent.”

read more SMPTE: Transparency and Education are Key to Governing AI

Bolin agrees: “It will be very intrusive at first and there will be some backlash. As the technology gets more advanced, users may stop seeing the ads as ads. But this will require a redefinition of TV advertising.”

The broad impact of AI/ML on advertising

AI/ML has the potential to revolutionize the way advertising works. Beyond personalized targeting, these technologies can enable more dynamic and immersive ad experiences. For example, AI can power interactive ads that adapt to user interactions in real time, as well as augmented reality (AR) and virtual reality (VR) experiences that create highly engaging brand interactions.

“AI can power interactive ads that adapt to user interactions in real time,” says Pike, “Furthermore, AI-driven analytics can provide deeper insights into campaign performance, enabling real-time optimization and more strategic decision-making. As AI/ML continues to evolve, we will see a shift toward more intelligent, engaging, and effective advertising.”

Bolin agrees, saying: “Ultimately, AI/ML can eliminate the concept of ‘ad breaks’ by replacing them entirely with content-integrated advertising.”

The integration of AI and ML into DAI has the potential to bring revolutionary advancements to advertising, but there are several open questions, including transparency, energy consumption, user privacy, etc. As AI/ML continues to evolve, the industry must navigate these challenges carefully to fully realize the benefits of intelligent, engaging, and effective advertising while deploying the technology responsibly and transparently.

read more IBC2024 is poised to lead innovation, explore trends and foster collaboration across the global media technology community.