FEED Autumn 2022 Newsletter

WATCH THE FULL VERSION OF THIS MASTERCLASS NOW

“WHEN WE USE THE TERM AI, MOST OFTEN WE’RE ACTUALLY REFERRING TO MACHINE LEARNING”

WATCH THIS VIDEO WITH QR CODE

NEAL ROMANEK: Do you see AI and machine learning being used in actual content creation?

BILL ADMANS, ATELIERE

MICHAEL PFITZNER: In the newsroom, we see automated news picking up. A lot of this is not really the creative part for journalists – it’s making an announcement on traffic conditions, or things that are repetitive and maybe only the data changes. Automated news, especially regionalised, is useful for offering content to areas where it would not be possible for a person to be creating. Then, we would leave serious content, of course, to journalists. You don’t want your articles to be fully written by AI, but automate some things. However, I can see it giving more information to people that normally would not have it. RAOUL COSPEN: Translation is an interesting use case in journalism – going from one language to the other, producing a first draft of a translation before it is handed over to be checked manually. We’ve been implementing that for captions.

find a specific person, talking on the phone, with a mountain on the right’. And it doesn’t happen. That’s why custom models are more to the point and give much better results. MICHAEL PFITZNER: It’s on us to find the right use cases and avoid misunderstandings about what an AI can and cannot do. Maybe there are expectations – expectations that you can replace all human work with AI, and it will function the same way or better, which is not the case. As Bill said, no training, no nothing – no knowledge, no recognition, no result. Sometimes you invest much more in training and still get inaccurate results, instead of using people like you used to. As vendors, we are here to consult with customers and tell them what is feasible. We should look out for creating expectations that are not fully reasonable. ANDREW BROADSTONE: In our space, people can come to it naively and think this machine is going to fix everything. That it’s sort of magic. It’s important that people understand ‘garbage in, garbage out’ still applies. You have to train the system, and there have to be patterns in the data in order to have anything meaningful. The user has to have a role in working with the system.

BILL ADMANS: One area that businesses like to use data is to make informed decisions. This is where the data gathered by AI can really help. At a macro level, when we look at the information we gathered from our CDN, it helps content owners understand what shows their customers are watching, what they should produce and acquire. Being able to fine-tune their content offering really helps maximise their ROI.

@feedzinesocial

Powered by