A new study has found that people prefer to watch a news video made by humans rather than one produced by artificial intelligence (AI).
Researchers at the Ludwig Maximilians University in Munich, Germany created three different news videos for topics about people like Donald Trump and Justin Bieber. One was automated by AI, one was partially automated with edits from human authors, and finally, one was exclusively made by humans.
In total, 42 videos were created and watched by 100 people who rated them and the researchers reached the conclusion that “human-made videos have the edge over those created with automation.”
According to Press Gazette, the audience preferred the human-made videos because they told them things they didn’t know, they were more professional and fact-filled, and were generally more engaging, comprehensive, and had a better story flow.
The audience judged that the AI-made news clips were more sensationalist and more emotional. Humans also made better use of audio, music, and photos.
“They [the AI videos] tended to be a little bit repetitive with the images that they used,” says Dr. Neil Thurman, who co-authored the research. “They didn’t do quite as well matching images to captions and that sort of thing.”
The AI also made mistakes, such as when making a video about Cristiano Ronaldo it erroneously included a photo of a different soccer player.
However, while there was a marked difference between a fully-automated video and a human-made video — the difference between a partially-automated video and a human-made one was not as significant suggesting that “a modest use of automation, with some post-automation human editing, can be well received.”
“One key takeaway of the study is that video automation output may be best when it comes in a hybrid form, meaning a human-machine collaboration,” adds co-author of the study Dr. Michael Koliska of Georgetown University.
“Such hybridity involves more human supervision, ensuring that automated video production maintains quality standards while taking advantage of computers’ strengths, such as speed and scale.”
While AI is sure to improve in the coming years, the research shows that for now machines still need human supervision in the video editing sphere.