Most technological trends see the light of the day and then recede, but the epoch of Artificial Intelligence in the M&E industry, is here to stay
By ANISHA GAKHAR
heodore: Well, you seem like a person, but you are just a voice in a computer.
Samantha: I can understand how the limited perspective of an inartificial mind might perceive it that way. You’ll get used to it.
Samantha: Was that funny?
Samantha: Oh good, I’m funny!
This excerpt is from an Oscar-winning American sci-fi romantic movie, Her, directed by Spike Jonze. The film chronicles the story of a man who develops camaraderie with an artificially intelligent computer bot, which is personified with a female voice.
Artificial intelligence (AI), also called machine intelligence or machine learning (ML), is scientifically devised intelligence demonstrated by machines, bots or computers, in paradox to the natural intelligence possessed by humans and other fauna. Advances in this sector, have enabled manufacturers and broadcasters to invent and implement smart tools and applications, to foster the quality of their content.
A good number of depictions follow suit, with movies like Resident Evil: The Final Chapter, Ghost in the Shell, Transformers: The Last Knight, Bladerunner et al, being bolted on the concept of AI. While there have been several instances of AI being used in international movies, it is yet to see the light of the day in India. An exception, is Rajnikanth’s 2010 flick, Enthiran, which garnered a lot of viewership and accolades in India. The story revolves around a robot named Chitti, with the ability to comprehend and exhibit human emotions, who falls in love with its creator’s girlfriend.
AI and its close relation machine learning systems, expand what can be made a recurrent computer task, thus freeing people to create more content and build audiences by finding new delivery routes, and ensuring optimum monetisation.
Glodina Lostanlen, SVP and GM – Americas P&N Sales & Services and Global Marketing, Imagine Communications, said, “What AI brings is the application of smart computing techniques to learn about a process, a workflow, or an operation, and automate it with increasing sophistication. The sorts of applications where AI is going to have a big impact are around advertising placement; for example, ensuring spots are seen by the right audience while optimising the use of the inventory.”
There was a time when AI was a niche area, accessible only to a handful of data scientists across the world. Adrish Bera, senior VP, OVP and Analytics, Prime Focus Technologies, said, “Today, open source based ML model builders like Tensorflow or Caffe can be used to build different types of Convolution Neural Networks (CNN) that mimic the brain’s ability to recognise complex objects and actions in layers. These model builders are computation intensive and can be hosted on any cloud-based virtual machine that has a powerful CPU and Graphics Processing Unit (GPU). “
According to Fabio Murra, SVP Product & Marketing, V-Nova, the company is currently experiencing an evident growth in the demand for high resolution and better quality streaming from consumers. Meeting that challenge is another area where AI can bring along a host of advantages. This enables efficient extrapolation of higher levels of detail from a lower base resolution, which helps to improve compression efficiency from contribution encoding, all the way through to distribution to consumers. This in turn may spell lower costs for operators and a better experience for viewers. Win-Win!
The Oscar-winning Her, chronicles the story of a human who develops camaraderie with a bot.
INTEGRATING AI WITH YOUR SOLUTION
Artificial Intelligence sits naturally in any software-defined workflow, optimising and ensuring efficient use of resources while maintaining quality output. AI components can be crafted as microservices as part of the architecture, and inherently virtualised, thus, allowing the AI system to be hosted in a data centre or in the cloud.
Accorind to Bera, an AI-enabled Media Asset Management (MAM) solution can help reduce the amount of effort and cost involved in cataloguing, managing, processing and re-purposing content. With such solutions, users can leverage AI techniques to generate meaningful metadata for maximising content discovery. This in turn helps them swiftly create compelling stories from vast content repositories, for documentaries, news stories, film promos and even live events. “AI integrations can also help automate localisation processes like subtitling and closed captioning, Additionally, AI can be leveraged to identify NSFW (Not Safe For Work) segments for compliance editing, helping editors locate shots depicting nudity, violence, smoking, drug abuse, and objectionable language,” he added. Other use cases include AI-assisted promo creation, and creation of sports highlights packages through automation in near real time.
According to Lostanlen, the Imagine Communications approach to the software-defined architecture is through microservices: small packages of code dedicated to one precisely targeted function, which can be permuted and combined in different ways to achieve varied tasks. Any workflow is built by pulling together the microservices one needs to get the job done, and then releasing them as soon as they are no longer needed. “We have an established platform called Zenium, which orchestrates not only our microservices, but also those from third party developers, allowing users to create precisely tailored functionality for their workflows,” added Lostanlen.
Murra said, “We have integrated AI into the core of our video codec PERSEUS. The implementation utilises in-loop Convolutional Neural Networks (CNNs) to train the encoder, in learning about common shapes and forms in video.”
Imagine Communications is working with Sky Italia, using its product Zenium, to enrich its compression and distribution operations using artificial intelligence. Because Zenium is an open platform, Sky Italia is creating its own AI algorithms to optimise and automate the allocation of resources. By introducing AI and machine learning into the content data flow, unique functionalies can be added. This enables reduction of operational costs, along with the re-assignment of personnel to creative tasks. Sky Italia sees the introduction of AI as a means of raising quality and saving costs.
This science finds its application largely in the M&E industry, in the categories of marketing and advertising, comprehension, search and classification, and experience innovation. According to medium.com, IBM used their AI system, Watson to help 20th Century Fox create a trailer for the horror movie Morgan. The AI system was taught to analyse and comprehend inputs from visual, audio, and other composition elements in 100 horror movies, so it could replicate moments during the making of the film. The six-minute movie trailer was completed in 24 hours, in contrast to a couple of weeks that it would have taken, had it been curated by humans.
Google, the leading search engine, avidly applies AI technology in its day-to-day executions. For example, the Google image search, enables users to upload sample pictures, and then uses image/face recognition to identify the image, or populate similar images. This spares the user the trouble of typing in keywords that relate to the image.
“We recently used AI technology for live tagging during cricket matches, using object identification, action detection, sound detection, score feed analysis and business logic to auto-tag the match footage and make it searchable within 20-30 seconds of the footage becoming available,” said Bera.
As user personalisation is the need of the day, companies are using AI to create personalised content for customers, based on recommendations that suits users’ personal tastes, whilst they are browsing or searching online. According to the Medium, Netflix’s content recommendation heightened when the company launched Meson, an intelligent workflow management application that recommends video based on users’ search trends.
AI IN NEWSROOMS
Imagine a world of bots wading through heaps of data, in order to generate news, in place of a journalist doing the same. The trend has dawned, and computers are sifting through lengths and breadths of databases, to facilitate automated reporting in newsrooms. According to the Business Standard, Google has provided Press Association, the British news agency, $8,05,000 to build Radar (Reporters And Data And Robots), that gathers, automates and writes nearly 30,000 local stories a month. Robotic Process Automation (RPA) is playing an instrumental role in automating several high-volume, highly transactional process functions, freeing up human labour to focus on more strategic work.
Media is an industry, where hoards of data are processed every single day in order to create content. The challenge though, lies in filtering out the fake and rumoured content, to ensure credibility and authenticity of the stories produced by the news channel. This not does not mean that bots will replace journalists any time soon, but as the media fraternity embraces the augmentation, the news business is expected to become highly automated, with AI-based tools serving as assistants at least, in the newsroom.
CAN AI WRITE A MOVIE?
How close are we to having computers pen down movies and content, like this article, for instance? Sunspring, the first short film to have been curated by AI technology in 2016, used text-recognition algorithms for the screenplays, as opposed to a traditional method of the writer laying down the script.
The novel, The Day a Computer Writes a Novel, which was almost wholly written by AI
, involved human intervention only as far as directing the plot was concerned. The sentences were generated entirely by AI. According to newatlas.com, the final outcome suggested that a degree of human oversight could make this kind of AI-written fiction, actually work wonders. It might be long before AI replaces human intuition and reasoning though, as we reasonably suspect.
ARTIFICIAL INTELLIGENCE, REAL BENEFITS
AI aided content creation facilitates umpteen possibilities like colour correction, automated editing, conformance/compliance and performance optimisation, scene extraction and highlights creation. Caption generation for photos and videos, language translations and automatic narration generation are also achievable through the use of AI technology. Face recognition through automated meta-data tagging and extraction, person-in-scene detection and person-in-picture tracking are features that can be leveraged, whilst using this augmentation. It also facilitates logo detection, text extraction and speech-to-text conversion. The person-in-scene tracking is especially useful during sports highlights, for instance, tracking the person who hit the goal during a football match.
“Applying recommendations to a content service would be revolutionary in content discovery and therefore in maximising revenue from a library. Like the move to IP and virtualisation, AI will prove to be effective when those that are engaged with the broadcast operations every day, learn about the technology, offer insight into how it can improve their roles, and become custodians of the technology. In short, AI is most effective when managed by human knowledge, intuition and insight,” affirmed Lostanlen.
WHAT THE FUTURE HOLDS
AI has alighted the M&E industry and how. The waves of this technology have started creating ripples, in arenas of films, video production, sports and broadcast. Lostanlen said, “The underlying principles of AI and ML have been known for 30 years. What makes it a powerful opportunity today is its ability to aggregate large numbers of processors while analysing a situation, and releasing them as soon as automated decisions are made, when implemented in virtualisation and the cloud.”
But before we step foot in this unparalleled era, there are a few pointers that need clarity. For instance, if AI were to create a piece of work, which entity would be authorised to legally own it? The costs associated with the set up of the technology, along with the manner in which it will impact the price of services in the industry, are some concerns that might come in the way of media houses wanting to embrace the trend. It might also render human involvement, a premium service, coming at its own heightened cost. Data privacy and security are concerns that imperatively need to be addressed before a company decides to execute an AI project.
But the pros perhaps outweigh the cons. Imagine a service provider that manages multiple television channels, each with its own balance of content types, audiences and value propositions. The company could use artificial intelligence to manage the scheduling, compression and delivery aspects through the different distribution outlets, and using a fixed delivery infrastructure. The result would be to achieve the optimum balance of quality, efficiency and value, while meeting audience expectations and contractual agreements. It will facilitate operational support, allowing human resources to focus on creating content and audience value.
How many miles the technology is set to cover, is unknown. Will it pause at content creation, or will it dive into review and strategy? Only time will tell. And maybe someday in near future, one of us might bump into an artificially intelligent robot with a network of artificial neurons, which may sound more humane than any human!