Beyond ChatGPT: AI is transforming video content for news media companies
Conference Blog | 03 October 2023
Although he is a filmmaker first and foremost, Dean Arnett has spent the past decade consulting with news media companies around the world.
For much of the past year, the former BBC producer has watched AI’s rapid evolution and began identifying some of the ways newsrooms can use it to become more efficient and create better content. That led him to develop the course AI tools for content creation to arm publishers with the right equipment for every stage of the creative process.
During the recent INMA Webinar AI and the future of editorial practices in Africa, Arnett gave a rundown of some tools he considers most valuable to news publishers. This is a key time for the industry, as these companies need to learn how to fit AI into existing workflows to improve their content and learn how to work with these tools.
Video’s new frontier
While many discussions about AI in the news media space revolve around ChatGPT, Arnett said he believes the image generation software DALL-E is what really made the public take note of what AI can do. Having overcome some of the challenges it had in the beginning — such as struggling with how to draw hands and feet — DALL-E has quickly mastered the art of, well, art.
But AI video generation is just beginning: “AI image generation has been around a while now and has progressed and evolved incredibly fast,” he said. “We are, however, at day one, hour one of AI video generation.”
Video-generated AI doesn’t require cameras, and people are already using it to make films and interactive art. And there are many tools newsrooms can leverage to transform video content. Tools are now available for every stage of the production workflow — from research and writing to archive management, content generation, music generation, image generation, video generation, video editing, post-production, and publication management.
“I get so excited about talking about AI,” Arnett said. “There’s so many tools that blow my mind, and I think there are so many tools that could have a day and night, night and day transformation in your workflows.”
AI can provide pre-editing of both audio and video, which will reduce time spent in post-production. Publishers can also streamline operations using AI for text-controlled video editing, auto-editing, re-versioning content, and publication management.
However, Arnett is quick to point out that AI will not replace an editing staff.
“What I’m saying is that with the intelligent utilisation of these tools, we can give AI the jobs which we don’t do properly [right now] because there is never enough time.”
For example, he said, making a promo for social media is usually done quickly because of a time crunch. Instead of allocating any human time to it, publishers could use AI tools that will take long-length content and create focused promos for social media platforms automatically and quickly.
“And not only that, these AI tools won’t [just] make one promo, which is what we do,” Arnett added. “AI tools will make different promos for the different social platforms knowing that Facebook is emotively led, Instagram is visually led, and X is thematically led. So these AI tools re-version our content for those platforms.”
To put the bow on that present, AI can handle the publication management to optimise SEO and suggest content descriptions.
Tools for content production
The tools available for AI video continue expanding and currently include avatar generators, voice enhancement tools, transcription translation and captioning tools, narration, generation and translation tools, and AI video-to-video tools.
One of the tools Arnett is impressed with is Rask, which clones voices and transcribes them, then can even translate them into another language. He shared a video he made with his sidekick, Vid Bot, using Rask:
The tools that can boost video content production also now include compositing, special effects, motion tracking, character replacement, and camera perspective generation. Camera perspective generation, Arnett explained, is fairly new, and several tools for it are in the early stages of release.
“It means that if I’ve got a [photo], let’s say I’ve got that classic shot of the Hindenburg in flames crashing, I can input that shot into AI and AI will then generate different camera angles of that historical event,” he said.
While that can create fascinating new imagery, it also creates an ethical issue that publishers need to consider now. And it isn’t the only issue: “AI is coming and it will change our newsrooms and our production workflow. I think we have a decision to make at this point, and ... the decision is, are we [going to be] subservient to our AI systems that are running everything, or are we going to take control of AI and make AI subservient to us?”
As tools become more impressive and allow for creating images and videos of people, places, and events that don’t exist, it will be incumbent upon news media companies to have guidelines in place to know how to manage them.
“There are a number of issues we need to be aware of,” Arnett said. “People say that AI’s going to be the end of the world. I don’t know about that, but it is going to have a big impact on society.”