Reuters is combining results packaging and synthetic media to create an AI-generated sportscaster onscreen. The synthetic figure is produced by a London-based AI company, Synthesia.
The AI-driven sportscaster is a digital twin of a real person, Reuters global sports editor Ossian Shine. His video image works the same way as a deepfake, digitally manipulating how the figure speaks, except in this system what’s said is drawn from real game results, retrieved and compiled with a Reuters in-house system. Early reviews say the presentation is convincing.
The result creates an automated workflow from field to screen. No scripting, no editing, and no production is required. Reuters calls their latest advance ‘the world’s first automated presenter-led video sport report.’ So far, it’s a proof-of-concept.
- COMPUTER BUSINESS REVIEW quotes the National Union of Journalists: “It is fascinating to see technology has progressed to the extent that robots can be TV presenters, but the personal touch and professional approach of a real journalist can’t be replaced. The technology should be used to enhance journalistic practices but should not be used as an opportunity to cut jobs.”
- FORBES quotes Synthesia CEO Victor Riparbelli: “I don’t see this as being an alternative or replacement for the six o’ clock news,” he says. “Rather I see this as a tool that enables broadcasters and companies to produce the video content that their audience demands.”
- PRESS GAZETTE quotes Ossian Shine: “It was a fascinating project to be involved in, but at the same time so surreal to see a character that looks and sounds just like you speaking phrases and sentences at someone, or something, else’s command.”
‘This kind of prototyping is helping us to understand how AI and synthetic media can be combined with our real-time feeds of photography and reporting to create whole new kinds of products and services.‘Nick Cohen, global head of product, core news services, in Reuters news release
- The finesse is putting a suite of AI systems in sync. In its most elemental form, it is one data set (the game results) driving another data set (the representation of the presenter’s face and speech). Visual highlight identification and automated writing systems work in tandem with an automated editing system, all feeding into an AI model that generates imagery of a speaking human.
- Similar systems could work in other areas, such as customer service, consumer how-to’s, online instruction, or others domains that communicate purely factual information.
- It remains to be seen if it could do a news report, but it is another step in that direction. FORBES says Reuters is open to ‘other areas besides sport.’
- Prototyping by a major western news agency likely will accelerate work on further possibilities.
- High degrees of personalization will be possible. Users could eventually select topics or languages of their choice, or even the nature and extent of details they wish to receive.
- Reports could be presented in multiple languages at low/no cost. This can open new markets at little cost, and in turn create added competition in those markets.
- So-called ‘AI anchors’ were first shown in China in November 2018, and there are now reported to be three AI anchors on-air. There’s another in Thailand, reports of two more in Russia, an another slated to debut in the UAE.
SEE RELATED STORIES
- Is AI Coming for the Job of your Favourite News Reader
COMPUTER BUSINESS REVIEW | February 11, 2020 | by Conor Reynolds
- Reuters creates prototype automated video match report led by AI sports presenter
PRESS GAZETTE | February 11, 2020 | by Freddy Mayhew
- Reuters Uses AI To Prototype First Ever Automated Video Reports
FORBES | February 07, 2020 | by Simon Chandler
- Reuters and Synthesia unveil AI prototype for automated video reports
REUTERS| February 7, 2020 | News release
Company website (with other examples)