We’re in an era in which emerging technologies, and especially artificial intelligence — where computers can perform the sorts of tasks generally thought to require human intelligence, and can “learn” and adapt — are becoming commonplace, and ever more capable.
Tools like OpenAI’s ChatGPT, which allows humans to converse with an AI and ask it to generate text — from cover letters and content strategies to haikus and news reports — hint at the disruption that’s coming for many content-dependent industries. ChatGPT’s contextual awareness means it’s able to create improbably realistic text with minimal human intervention, and it acutely demonstrates the upheaval to come.
AI is impacting the media and entertainment industries in several ways – from content creation, personalisation and metadata tagging to controlling how online content is distributed, how content is categorised and classified, and identifying bogus content. Similarly, AI is enabling automated journalism, from performing research for reporters to automatically summarising and aggregating articles, acting as an editor, and even standing in as a virtual news anchor.
McKinsey estimates the potential total annual value of Al and analytics across industries in 2025 will be between US$9.5-trillion and $15.4-trillion. Of that, it expects around $448-billion in value will be added to the media and entertainment industries by AI. Meanwhile, PwC estimates the global entertainment and media sector will enjoy a 5% compound annual growth rate that will raise industry revenues to $2.6-trillion in 2025, up from $2-trillion in 2020.
It’s not only what’s possible with AI that’s changing, but how people interact with businesses, use their services, and what they expect from them. At the same time, it’s helping media and entertainment businesses improve their offerings and better cater to their customers’ evolving needs and expectations.
Metadata tagging
New videos, still images, illustrations and other content are uploaded to a wide variety of platforms every minute. That makes categorising and classifying all this content an enormous task for companies, but it’s a task AI can do with ease. CBS Interactive, for instance, uses AI-powered video analysis tools to examine every frame of a video and tag objects in it. The US National Football League is using similar technology to help it structure and organise its content, making it easier to search, use and monetise as a result.
Content personalisation
If you’ve ever used a streaming service like Netflix or Spotify, or been served an online advertisement for a service in your neighbourhood, you’ve seen content personalisation at work. On-demand services use AI and machine learning in conjunction with any demographic or behavioural information they can glean about users to try to suggest other content in their catalogues that will appeal to users and keep them engaged with the platform.
Reporting automation
Streamers and broadcasters alike need to keep track of how content is performing and adjust their strategies accordingly. Traditionally, that’s meant enormous spreadsheets and a great deal of human intervention on an ongoing basis to track performance. Now AI can not only create these sorts of reports from raw data, but in some instances can make suggestions on how to harness the insights they reveal.
Subtitle generation
International content creators and distributors often need to adapt their content for different markets, which can include creating multilingual subtitles. Doing so with human translators and subtitle creators is costly and time consuming, but AI is making it much easier. Already, YouTube’s AI makes it possible to automatically add closed captions to videos. The technology is set to help businesses find new efficiencies while also cutting costs and improving users’ experience.
Read: Microsoft to add ChatGPT to Azure cloud services
Search optimisation
AI is changing how search works by, for instance, making it possible to search for a particular image, or in the case of Google Lens and similar services, to use a smartphone camera pointed at an item in the real world to initiate a search for the same product at an online retailer. It’s also making it easier for online retailers to deliver the right result to users when they initiate a search, improving the odds they complete a transaction rather than get frustrated and take their business elsewhere.
VR, AR and chatbots
Virtual reality (VR) and augmented reality (AR) are becoming more widespread, and consumers are becoming more comfortable using them. AI can assist in the creation of VR and AR content, which drastically reduces production costs and means a growing number of creators and businesses will be able to embrace the technologies. Similarly, as conversational AI improves, so does its ability to provide customer service solutions, like Google’s service that can call a restaurant or service provider and make an appointment on a user’s behalf using a virtual voice that’s able to react realistically to the person on the other end of the call.
Managing bogus content
AI can help with content moderation and access to it by, for instance, detecting a user’s age. It can also help appropriately categorise content, as in the case of True Anthem AI, whether for enforcing age restrictions or simply making substantial catalogues easier for users to navigate. Then there’s the growing challenge of identifying erroneous or fake content, something AI is particularly adept at, because it can extrapolate from massive data sets, fact-check information, and update its assessments in nearly real time based on new data.
Journalism and media
Media companies rely on a range of roles and tools to create content for various platforms, and vary in size from local outlets with limited resources to global enterprises. In the same way, AI can help with fact-checking or flagging information deliberately designed to mislead, it can also help journalists identify and pursue the stories most important to their audiences.
AI can help journalists with their research and information gathering, even if they don’t speak the language. Connexun’s AI tool can extract content from articles around the world and in multiple languages, making it easier to identify trending topics in the global media.
AI is also able to check the veracity of journalists’ stories, assess their stories’ performance, draw insights from them, arrange and archive content for future use, and repurpose it for different channels. Design tool Canva now supports text prompts, making it easy to create supporting illustrations or infographics for stories.
Google Docs, meanwhile, can now create AI-generated summaries of articles. In general terms, there are two computational approaches to AI-based summary creation: extractive and abstractive. In the case of extractive summaries, the AI pulls out what it deems pertinent pieces and stitches them together. In the case of abstractive, it aims to use the input document to create a cohesive, factually correct and grammatically sensible summary.
According to Futuremarketinsights, the global market for AI in media and entertainment was valued at $10.4-billion, but it’s estimated to grow at a compound annual rate of 26% from 2022 to 2032, reaching a staggering $132.2-billion. It’s easy to understand why when you consider products and services like Focal Points.
Focal Points leverages the power of machine learning and natural language processing (NLP) to sift through large quantities of data — including media coverage and client information — from which analysts can then compile easy to digest reports.
Read: Is ChatGPT the start of the AI revolution?
“During the analysis process, it is important to identify and separate what volume of the media coverage was produced from ‘owned’ or ‘client-specific’ content. This includes press releases or product specifications from journalists’ opinions or recommendations,” says April Parry, internal media analyst at Focal Points.
Automated storytelling
If all AI did was make it easier to translate data into stories in plain language, that would be remarkable, because doing so allows for data-driven decision making, and it allows for data-based storytelling at a scale that’s impossible using human effort alone, simply because there is so much data to contend with, and it’s growing all the time. Fortunately, AI does this already, and far more besides.
Automated storytelling tools use AI and machine learning to create content fit for human consumption. It can be used to create headlines, summarise businesses’ year-end financial reports, or — using tools like natural language understanding (NLU), which can turn unstructured data into narratives — it can even do more traditionally creative narrative pursuits like writing poetry, short stories or screenplays. In some instances, human intervention is needed, but automation can provide a solid starting point, or encourage fresh ideas.
In the case of number-heavy narratives — like post-game reports on statistic-laden sports — it can often do the entire job. The Washington Post, for example, has used a tool called Heliograf to write over 850 articles. Heliograf has been used to cover the Olympic Games, US elections and even high school football matches. Reuters, meanwhile, uses a tool called Graphiq to provide news outlets with free, interactive data visualisations that update in real time.
An emerging example of the field is the company Narrative Science, which uses data storytelling and natural language generating (NLG) combined to turn research data into easy-to-comprehend narratives. It also builds dashboards for users that help them extract useful information from data. One such service, called Quill, can plug into existing solutions like Qlik, Tableau and Power BI.
Meanwhile, the first movie or play written wholly by AI can’t be far away. Researchers at the University of Vermont have developed an ML tool that identifies emotional arcs in texts or videos … and then creates arcs of its own. A similar tool from McKinsey’s Consumer Tech and Media team can trawl videos and online content and derive sentiment analysis from the plot, choice of camera shots, music and dialogue.
AI also frees up journalists to do more interesting work. The Associated Press uses a tool called Wordsmith in conjunction with earnings data from Zacks Investment Research to turn quarterly results reports into publishable stories in seconds. It uses the same tool to report on basketball games, leaving journalists more time to write more in-depth and qualitative pieces.
Read: Artificial intelligence is Lenin’s rope – a totalitarian’s dream
If you’ve ever visited the comments beneath a story on the New York Times website, you may have noticed it doesn’t present them chronologically, but instead surfaces those deemed most relevant. The arbiter in this ranking process isn’t the outlet’s editors, nor is it the result of users up- or down-voting posts. Instead, it’s thanks to a pair of AI tools: Editor and Perspective API, which moderate and sort user comments so the most useful appear first.
Chatbot media interfaces
Beyond customer support, chatbots have many potential uses in the media and entertainment industries. They can help users surface pertinent information, like archival stories on particular topics, or current stories in their areas of interest. Chatbots can also assist with checking account details, subscriptions and profile settings, and can help users update payment information — something that can be key to keeping subscriptions consistent. Increasingly, they’re also being used for gamification and other systems that help keep users engaged.
Quartz has been experimenting with an app that more closely resembles a chatbot than it does a traditional news app. The aim, of course, is to surface more content a user will care about and engage with, and ensure that those topics continue to appear and are presented in the way the user prefers — that can mean smart speakers, mobile devices or even connected devices in vehicles.
The company has also created the Quartz AI Studio, which helps journalists create stories with the assistance of ML, whether it’s sifting through large data sets, summarising content or repurposing it for different channels. Quartz hopes its solutions might empower smaller newsrooms to produce the sorts of stories that would otherwise be impossible by reducing the burdens on journalists without compromising the quality of their stories.
AI and the news
The state media agency Xinhau in China has tested an AI news anchor that uses “multi-modal recognition and synthesis, facial recognition, and animation and transfer learning” to imitate human mannerisms and voices. Similarly, a company called MoneyBrain makes AI avatars that many South Korean broadcasters are using to replicate existing anchors.
While artificial anchors may seem like a threat to the news sector, AI’s ability to detect fake news with unprecedented accuracy could help defend it. AI is far more proficient at recognising fake news and deepfake content than humans are. Services like Fabula AI, which was acquired by Twitter in 2019, use algorithms to identify the patterns that immerge from fake news spreading online. Another service, Grover, is a fake news detection AI model from researchers at the University of Washington that focuses on language use.
Read: Curro pilots artificial intelligence for learning in its schools
These tools allow journalists to refocus their attention on the stories that matter. As such, AI is helping to reimagine the role and potential scope of human journalism. “It frees our journalists from routine tasks, to do higher-level work that uses their creativity; allows us to create more content that serves new audiences more efficiently; and improves our ability to discover news,” says Lisa Gibbs, the director of news partnerships and AI news lead at the Associated Press.
That’s not to say AI will replace journalists entirely. AI tools in use at Bloomberg flag trending topics and collate data, but humans still ensure the core details are correct before creating stories from AI’s leads. The need for human minds, capable of making complex emotional connections, judgment calls and contextually sensitive deep dives isn’t going anywhere any time soon.
Read: SA government to establish artificial intelligence institute
Sometimes the limitations of synthetic solutions are more pedestrian. Xinhua’s AI anchor once referred to Alibaba founder Jack Ma as “Jack Massachusetts”. An honest mistake for an AI, but one a human is unlikely to make.
What’s more likely, as Nicholas Diakopoulos suggests in his book Automating the News: How Algorithms are Rewriting the Media, is that newsrooms will remain full of people, but they’ll be doing slightly different things as their roles evolve. Humans will likely work alongside AI, making the most of its abilities while continuing to compensate for its inherent limitations.
- The author, Dr Mark Nasila, is chief analytics officer in First National Bank’s chief risk office