Consulting & Innovation
Solutions & Technologies
Infrastructure & Operations
Industries
More
Blog
Bild_Medienbild

AI in Content Production

How AI helps distinguish between hundreds of celebrities

AI in Content Production
28.03.2019
Cloud
Media & Entertainment


Artificial intelligence (AI) is not just buzzing through the media as a buzzword. It has also arrived in media production. The application scenarios for artificial intelligence are becoming increasingly diverse.

How AI helps distinguish between hundreds of celebrities

Imagine you work at a TV station. The Oscars are coming up in L.A. and your job is to report on this major event.


However, you have a problem: You don't even know many of the American stars who are currently strolling down the red carpet. You're supposed to cut together short reports about individual stars to supplement the evening's moderation.

Software solutions based on AI - like Azure (Microsoft) services - can save you in this situation. The company's cloud-based applications are able to match the latest clips from your camera crew with a face database stored in the cloud. With a 99.9% probability, the system tells you which star is currently being interviewed by the presenters.


At the same time, the software shows you if there are already other clips with this star in your audiovisual library. You can incorporate these clips directly into your own reporting.

Examples like this illustrate how AI is already helping to significantly simplify the work of a TV editor.

Narrow AI - Small but mighty!

Nevertheless, reporting on artificial intelligence is often fraught with skepticism: It often deals with job losses, data misuse and ethical dangers arising from the use of the new technologies. Unfortunately, this form of presentation neglects the current possibilities and limits of the use of artificial intelligence.

Current systems, which are summarized under "AI" (Artificial Intelligence), are completely assigned to the area of Narrow AI. Narrow AI refers to algorithms or software that specialize in one activity and continue to improve this activity through learning processes. Thus, these Narrow AI systems can be described as comparatively "dumb". They master activity A, but have to pass when it comes to activity B.

What Narrow-AIs can do - and what they can't

Narrow AIs' capabilities range from speech recognition, data visualization, and storytelling to independent creation of images and musical pieces. Through training, narrow AIs can achieve performance levels that exceed human capabilities when dealing with specific problems. For example, algorithms for traffic sign recognition are already being used in automobiles today, far outperforming experienced drivers in terms of accuracy and speed of identifying signs in traffic.


Narrow AI systems are still far from performing human and creative activities. Rather, they are utilities that can support humans in their daily work.


Media companies can benefit from these tools to drive their own content creation.


Application scenarios of Narrow AI along the work processes of media companies

Narrow AI systems can be used in all work processes of a media company. In this way, the entire value chain benefits from the relief of its employees. This will be illustrated by selected examples:


  1. AI-based programs are already helping editors create content. Text-based robot journalists are no longer the only ones in use to create stock market, sports and weather news. Software, such as the company's wibbitz video editing software, automatically creates news clips from data provided by news outlets.
  2. AI-powered dashboards like Brandwatch report which topics and trends are currently being discussed in social media channels. In this way, these systems help editors maintain an overview of the daily flood of data.
  3. Once the "first copy" of the media product has been created, AI-based cloud systems from providers such as Amazon or Microsoft help control the playout of huge amounts of data that an online provider such as Netflix streams to end customers every second.
  4. Companies like Spotify use AI-powered systems to store their customers' usage habits. The data is used to suggest the right media products to each user in a targeted, individualized way that matches their current mood.
  5. In terms of advertising placement, AI-based ad servers are already demonstrating how to select advertising space for specific target groups. In the coming years, these principles will gradually be transferred to other media, e.g., to so-called addressable TV. For example, AI-based real-time advertising systems are already able to recognize individual customers and their preferences on the basis of stored information (using so-called cookies) and play out the appropriate advertising messages.


And how are these systems being used in media companies? - The API and PaaS revolution

More than 20 years ago, the founder of Internet giant Amazon wrote an internal memo (called the API Manifesto) that radically changed the way his company worked. At the time, the API Manifesto anticipated a development that is now considered one of the most significant revolutions in the use of software in companies.


The abbreviation API stands for "Application Programming Interface" and can be translated as programming interface in German. APIs make it child's play to integrate different applications into your own software.


Narrow-AI to go, please!

This will be illustrated with an example: So-called speech-to-text applications are an important narrow AI system. In recent years, these have achieved a level of performance that also enables professional use in media production. These algorithms are capable of converting spoken text into written text in real time. They are also used by Apple's Siri or Google Now, among others.


In order to use speech-to-text in their own editorial offices, no media company needs to program such a system itself anymore. Instead, it is sufficient for the IT department to enable a data transfer via API to the systems offered online by the major manufacturers for the employees in the editorial offices. The editor can then upload his audio content at the touch of a button. Within a very short time, he receives back a text file that he can incorporate directly into his reporting. And he can do so in the text format that he has set beforehand.


AI helps professionalize work processes if you let it

AI systems from various providers are already capable of supporting the daily work of editorial teams in publishing houses and broadcasters. This is a great opportunity - especially for German media companies. Even with a small initial investment, media companies can professionalize their own work processes - and shorten the gap to their American competitors. From content creation to sales and customer contact.


The bigger problem is therefore not the technical feasibility of AI application scenarios. Rather, it lies in the reservations and intuitive rejection of many editorial offices and media houses to use such systems in daily operations.


AI in Practice

Be inspired by our application examples for AI. Discover our AI cases from practice.

Arvato Systems AI Services

We bring artificial intelligence to your business. Discover our services and solutions around AI here.

Our solutions for Media & Entertainment

Get an overview of our entire portfolio for the Broadcast & New TV market.

Written by

MA_Kathrin_Kleinschnittger_Cloud
Prof. Dr. Roland Frank
Professor Mediadesign Hochschule