This article has been taken from IABM Journal Issue 109.
Director Strategic Insight, IABM
In a world currently dominated by discussions around the transition from point-to-point connections of video and audio to network connections over IP, we may have overlooked that this is not the end game but rather a stepping stone to something bigger.
Networked media operations open up access to resources and processing power that were just not available in the proprietary world of fixed non industry standard (IT) connections. Therefore the huge resources and rapid developments occurring in those areas could only be partially exploited. The obvious example here is access to Cloud storage and compute.
But increasingly those barriers are being eroded and new opportunities are emerging, and the Cloud with a range of processing and management tools is an important component. Most of the Cloud providers have AI tools on offer as services coupled to on-demand storage and processing power. This is a real gamechanger. The IABM report on ‘Media Tech Trends – Artificial Intelligence’ highlights an already growing interest, as shown in the adoption chart.
Even if the Cloud isn’t on your radar, the world of IT offers on-site technology that can revolutionize operations by re-engineering efficiency, opening up new programming opportunities, changing how on-line providers service consumers and redefining the skills needed to develop and maintain the resources.
There is no need to be overly bothered by the semantics and differences between AI and machine learning. True they are often wrongly used interchangeably, but collectively they contribute to the trends outlined in this article. Strictly speaking AI is the broad science of mimicking human abilities, while machine learning is a specific subset of AI that trains a machine how to learn from experience.
Media and broadcast is a very broad discipline, ranging from content production to consumer interaction plus the orchestration of everything between. So what AI means to each subset of the journey from camera to viewer is totally different. We can however neatly categorize the different aspects using the IABM BaM Content Chain® which is built around the following creative, technical and business processes: Create, Produce, Manage, Publish, Monetize, Consume, Connect, Store, Support.
Applications of AI
For content production which covers the Create and Produce categories of the Content Chain, projects are underway to enhance creativity and also to produce more content with reduced resources. This has led to controversial discussions around the potential for AI systems to replace creative activities. Creativity is seen as a precious human skill and a capability which, if replaced by software and processors, is considered by some a threat to our very existence. This is not the right way to view these developments.
The use of AI and machine learning to automate camera production is a huge opportunity to bring new content to viewers that would otherwise be prohibitively expensive to produce.
Obvious examples are sports events and music concerts, and even drama is a potential candidate in the longer term. Current experimental productions have shown that while a human can make editorial decisions which trump what automation can do, the gap will narrow but probably not close completely.
Nonetheless, lower cost automated production is a gamechanger.
Audiences’ expectations have developed over the years; they appreciate the ‘hand crafted’ skill of a high budget production but they also appreciate access to events and programming that reach good (but not the best) production values, if the alternative is no access at all. So AI and machine learning can help meet the need for more content at reduced cost.
Importantly though, all those in production roles will acknowledge that the process is very labor intensive and not all of it creative; anything that can be done to optimize the production process and support the human creative process is positive. It will almost certainly lead to an even higher level of innovation and job satisfaction.
Orchestration is a word increasingly used across the Content Chain, particularly in areas such as Manage, Connect, Store and Support. With systems becoming increasingly complex, the challenge of operating them efficiently is growing exponentially.
Orchestration is defined as the automated configuration, coordination, and management of computer systems and software. From this definition we can immediately see the potential for AI.
These are multi-faceted activities and highly repetitive. In many cases, the decisions made influence costs, performance and reliability. These activities lend themselves to automation driven by artificial intelligence decisions. Importantly though, they can still alert operators to the most important issues requiring their skill, decisions and expertise by separating these out from the mass of lower order issues which can be taken care of automatically.
[bctt tweet = “Data is the fuel that drives all forms of Artificial Intelligence and automation. Collection and use of many types of data quite rightly needs careful control or it becomes an intrusion on privacy or worse, opens the door to serious malicious practices and theft. – AI Article in IABM Journal”]
Optimization is an important element of orchestration and in a virtualized world with software defined networks, the potential for intelligent automated operations is clear. Decisions have to be taken to ensure resources are used efficiently, are available when required and produce reliable, consistent performance.
Finally, we can envisage the growing need to manage content, making intelligent decisions about finding content with specific criteria, and decisions about storage based upon anticipated content usage –for example, to make content available instantly on-line or archive it to less accessible deep storage.
Engaging consumers with Publishing, Monetization and Consumption is where the money comes from to power the entire industry. Finding the keys that unlock consumer engagement is a prime role for Artificial Intelligence. Every consumer can be treated uniquely as an individual –something that is impossible without this kind of support.
As the Internet of Things becomes a reality and with products such as Alexa becoming ubiquitous, consumers expect a personalized service that understands their needs, interests and aspirations. Unfortunately they don’t want to pay for it directly, so this new intelligence needs to provide a service to consumers they value and at the same time maximize the monetization opportunities.
The good news is that directly or indirectly there’s plenty of source information to analyze and to derive insights from. So with the right AI tools and enough processing power, the potential is there to redefine the user experience.
Security – threat, help or hindrance?
One aspect that complicates progress is security, with several very high profile breaches and abuses of personal data demonstrating the challenge.
Data is the fuel that drives all forms of Artificial Intelligence and automation. Collection and use of many types of data quite rightly needs careful control or it becomes an intrusion on privacy or worse, opens the door to serious malicious practices and theft.
There is no doubt that while progress is slowed by the need to respect data protection, it is not stalled; the need to develop new tools for more sophisticated and complex operations is too compelling.
So, are we underestimating the role of AI in Media and Broadcasting?
Almost certainly yes. Without rapid development of AI and machine learning techniques applied to broadcast and media, our ability to manage facilities and grow businesses will be severely limited as complexity increases.
In many areas the use of AI can and will do a better and more consistent job than even skilled staff. Increasing headcount is the only alternative and that can be self-defeating: doing so increases complexity and the scope for errors especially as we are all different in our strengths and weaknesses.
We will see several examples of spectacular AI failures over next few years and claims they are the result of misplaced faith in the technology, but this is all part of the learning curve and not an indication of a flawed strategy. With every new technology, especially in the early years, there are examples of mature legacy systems doing a better job than the eventual successor.
In traditional broadcast engineering circles, there is a lot of debate and excitement around the move to IP interconnects. This in itself has a lot of legacy thinking, with the emphasis on point to point connections. In reality, it provides a stepping stone to massively increased use of virtual machines, Cloud processing and AI, which in turn opens the door to an exponential increase in processing and storage capacity. Point-to-point connections will become less important as different interconnected steps communicate within the confines of the processing environment.
What about innovation, creativity and the role of skilled operators?The good news is that, with the availability of more sophisticated tools, they will all evolve to a higher level and create even more compelling entertainment and solutions.
In the workplace, the skills needed are already changing and this is a huge challenge for traditional enterprises. Transformation is required at a much more comprehensive level than a change of technology. It’s a sad fact that more attempts at radical transformation fail than succeed because restructuring the skills and work practices of individuals is substantially harder than replacing hardware and software.
The future is bright but very different and remember what we are experiencing right now is the ‘stepping stone’ to the future, not the end game!
For a more detailed analysis of the current status of AI and its deployment, view the IABM report ‘IABM Media Tech Trends –Artificial Intelligence’