- Integrated Systems
- About Us
- Integrated Systems
- About Us
Deep Inside the Mind of Media
Three takeaways from the recent HPE & NVIDIA conference on AI in Broadcasting
On Thursday 28th November HPE & NVIDIA hosted more than 50 people from across the Media sector to get to grips with the impact Artificial Intelligence will have. It was great to see such a large audience for the event and high levels of interest around AI. Throughout the afternoon we had an unprecedented number of questions across a broad variety of topics. It became clear that most of companies represented in the audience feel they are only at the beginning of their AI journey. However, with so much hype and promise around AI a lot of people are keen to get down to the nitty-gritty, which means asking the questions like; where are we with AI today and how can it help me?
So to break it down for you, here are some of my key takeaways from the discussion:
Setting the Scene for AI in the Media Sector
AI today can tackle two main areas; creation of content and analytics & indexation of existing content, with creation having the greater promise but being the more challenging to achieve. Even in the analytics space AI today is far from changing complex end to end business processes. AI is definitely breaking into the creativity space, but it is early days. To that end we heard from two startups who are working in this area; DeepZen who are taking voice to the next level and Synthesia who are doing ethical things with Deep Fake technology. In terms of gaining insights or monetarizing large volumes of data, something the industry definitely has, AI is an essential tool. AI should be an instrument in an organisations’ toolkit that can be used to address current and future business challenges as well as create business opportunities. Any AI implementation should be based on a solid strategy with specific use cases evaluated and selected based on their ability to drive the right value. AI is already revolutionizing other industries by helping organisations understanding their customers, optimizing time and automate their back office and the media sector should be no exception.
Changing Customer Demands
In recent years consumer behaviour towards digital channels has changed massively, expectations are at an all time high. Highly personalized content is critical to attract and retain audiences in a highly competitive market. Consumption has now gone digital, meaning a much larger stress on computational power. High Performance Computing combined with the acceleration technology in GPUs has been critical in the establishment of AI due to the levels of data ingest required. So ultimately, technology has caught up with the theory, enabling us to make quick changes in the industry.
AI is being used to give a personalized experience, curate content and target title recommendations all at scale. Implementation of this technology is critical to gain insight into audiences, their preferences and how to maximize both the viewing experience and also their loyalty. Thus driving a significantly better return.
This is best achieved when combined with data obtained from video metadata, including timecoded appearances of actors, what, when & how they are speaking, the objects they are interacting with and how brands are placed, This builds towards a more intimate relationship with the consumer and helping them find, and have pushed, the right content and marketing.
Where else can AI be used in the Media sector?
There is huge amount of opportunity for AI to play a significant role in transforming the sector, from understanding user preferences to speeding up the creative processes. Check out the innovative GauGAN from NVIDIA which uses Generative Adversarial Networks (GANs) to speed up the creativity process by turning doodles into photorealistic masterpieces.
AI can also add value to the media value chain by; easing any mundane burdens to increase productivity from creation to editorial, predicting the demand, allowing dynamic adjustment in production, fixing delays in the supply chain or using the technology from Valossa to ensure compliance. The significant possibility is that AI will be at the forefront of creativity. Using these techniques will allow near real time analysis of the global consumer base to bring content consumers into the loop with Artists in a much more dynamic way.
There are many outstanding questions about the role of ethics and intellectual property in the space, but the technology is here and available. Adoption of AI in the Media Sector is not a matter of when, it is already happening and companies that are at the forefront will be watching their laggard competitors fall by the wayside in the near future.
What is the Future for AI?
We are also starting to see a new breed of Software Development Kits being released that allow multiple AIs to be combined to drive a more comprehensive outcome. Jarvis from NVIDIA is a great example of this, fusing vision, speech and other sensors it creates a multi-modal AI or collection of AIs that are working together. This technology allows complex workflows to be constructed and GPU-accelerated, for example understanding and utilise visual cues such as gestures and gaze along with speech to create responses that are much more contextually aligned to the situation.
With much of the content creation process happening outside of the traditionally centralised data centre or cloud, we are also seeing a proliferation in the need to gain insight into rich media at its inception, where it is created - at the edge. Using a highly distributed framework, Swarm Learning from HPE is a decentralized AI that operates near to these distributed data sources. This is where data is most fresh. In this completely decentralized architecture, only learned insights instead of the raw data is shared among the collaborators in the swarm, which tremendously enhances the ability of the model, data security and privacy.
New to AI?
If you are new to AI, check out my other articles on the HPE UK&I blog in which I cover the basic of what AI is and also how it works. At HPE our HPE PointNext Services use a tried and tested method of Explore, Experiment & Evolve and have helped customers start identifying use cases or continue on the AI journey. We can also help you get to grips with this transformative technology through our partnership with the NVIDIA Deep Learning Institute and whether you are starting small or already running complex neural networks we have the technology.
Hewlett Packard Enterprise