Researchers say automated news video production is better with human intervention

AI-generated short message videos will only be rated as well as manually created videos if they are edited by a human.

News organizations such as Bloomberg, Reuters, and The Economist have turned to AI-powered video services to meet the growing audience demand for audiovisual material. A study recently published in the journal Journalism shows that automated production of news videos is more effective when under human supervision.

Technology providers like Wochit and Moovly allow publishers to mass-produce videos at scale. But what will viewers think of the results? The researchers, led by LMU communication scientist Professor Neil Thurman, found that only automated videos with human post-editing performed as well as fully human-produced videos. I discovered that I like it.The works are published below journalism.

“Our research shows that news consumers, on average, prefer short-form automated news videos just as much as manually produced news videos, as long as the automated process includes human oversight. ” said Neil Thurman of LMU's Office of Media and Communications.

Mr. Thurman, along with Dr. Sally Starrs (London School of Economics) and Dr. Michael Koliska (Georgetown University), will develop human-produced, highly automated, and partially automated videos covering a variety of topics. We evaluated the reactions of 4,200 UK news consumers to Topics include Cristiano Ronaldo, Donald Trump, Wimbledon Tennis Championships and more. The partially automated videos were post-edited by humans after the initial automation process.

The results show that there was no significant difference in how much news viewers liked the human-generated videos and the partially automated videos overall. In contrast, the number of likes for highly automated videos decreased significantly. In other words, the results show that automation of news videos is better than human monitoring.

According to Thurman, “One of the key takeaways from this study is that the output of video automation may be best suited for hybrid formats, i.e. human-machine collaboration. Hybridization involves more human supervision, allowing automated video production to maintain quality standards while leveraging computer strengths such as speed and scale. ”