Generative AI continues to drive conversations in virtually every industry as stakeholders look at how they fit in this changing landscape. This week’s Webinar gave INMA members a glimpse of what challenges news publishers face around licencing content data and how to prepare for them.
Virl Hill, the former head of worldwide business development and strategy, media, and entertainment at Microsoft, joined INMA’s Ariane Bernard, lead of the Smart Data Initiative, to discuss the risks and opportunities associated with generative AI. During Navigating IP, licensing, and the opportunities of generative AI, Hill answered members’ questions that had been submitted in advance.
Bernard noted that the lack of regulations around AI raises many questions about fair use and how publishers manage their IP:
“We have to think about what IP means for us. We have to think about our value proposition to these platforms, but also about what makes us unique so that we licence what can be a commodity that has value for ourselves and others,” she said.
At the same time, it’s important to retain some control over that IP: “And all of that kind of leads us to questions around pricing and deal terms and also questions like privacy and compliance.”
As companies approach AI and look at how to leverage it, Hill urged them to keep three important points in mind:
- AI represents a “profound opportunity” that already is democratising content creation. “Generative AI can be used as a tool and it can be used as a creation device,” he said. It’s also a productivity tool that can be incredibly beneficial organisations — and Hill anticipates it will “create new types of media that we haven't even envisioned yet.”
- Be prepared. This could include forming an internal AI standards board or council. “This is an opportunity for you all to help shape and define. Don’t look at this as something where Big Tech is coming to you and saying, ‘This is what you have to do.’ You have the inputs that shape the outputs of these systems,” he said. “So have a stake at the table, have a say in how it goes.”
- Humans created AI, and people are responsible for ensuring it is used appropriately and responsibly. As organisations think about AI within their company, Hill encouraged members think about what standard of responsible behaviour their organisation will maintain. “Set your own standards for what’s safe and what’s appropriate and be transparent,” he suggested.
An era of creativity?
Rather than being a takeover by technology, this could be the start of a new era of creativity, Hill said. Generative AI models will continue improving, and the creative output — such as computer graphics and animation — will get more realistic. He predicted natural language processing will be vastly improved, especially in things like chatbots, virtual assistants, and other types of Q&A environments.
While he doesn’t expect it to replace people in tasks like writing news stories or content, he does see it becoming an even better assistant for researching stories.
“Generative AI can take away some of the redundant or difficult tasks, the research-driven tasks, but it’s not going to replace human creativity,” he said. “Human beings are going to have a profound opportunity to build on top of it in the next two to three years. And I think that’s the most important thing you’ll see.”
Licencing and IP
One question that is already in the minds of publishers is the licencing angle. Bernard questioned how publishers can establish the value of their content data and how it can be protected —particularly since generative AI will essentially take whatever information it has scraped and keep learning from it.
“Once those AI platforms learn from content, they don’t unlearn it,” she said. “It’d be really hard just to say, ‘I want to undo the connections that were created from crawling this content.’ So the licence is permanent.”
That differs from other arrangements where publishers can licence content to a vendor but then recall it when the licence or partnership ends. This will require creating a new business model, Hill said.
“Using training models is not media consumption in the classic traditional way,” he said, comparing it to attending college: “If I go to college, I buy a bunch of books … I sit in lecture halls, and I am getting intellectual property from my professors. I am getting from them the vast wealth of their knowledge presented to me, and I’m going to apply that knowledge for the rest of my life, hopefully, in doing what I do.”
The college doesn’t retain the rights to that knowledge and it has been paid for providing it. Rather than looking at old licencing models, Hill said publishers should consider where and how they want their content to be leveraged and deployed.
“It’s a business decision at the end of the day as well as a philosophic decision: Is my content important to that customer base? I don’t want to be there just for the sake of being there,” he said. “So I would start from the valuation perspective of asking myself, why am I here and is this important to me?”
If it is deemed important, then companies must determine what it’s worth. That will be a different model than consumption-based licencing for subscribers, and if the deal isn’t attractive to the publisher, they don’t have to agree to it.
In these early days of AI, one challenge is that there’s no measurement yet of what kind of compensation publishers should receive for their IP. So there’s a watch-and-wait mentality, with Hill predicting many news media companies will sit on the sidelines and watch to see what happens before hammering out their own licencing and IP strategies with AI.
But even if they’re not ready to make a move, every news media organisation should be developing its internal AI strategy and looking at how it relates to the overarching company initiatives.
Preparing internal teams
Companies must also ensure that internal teams are properly aligned to support the strategy. Otherwise, “you can be the greatest biz dev person in the world and go out and negotiate a hell of a deal, but it’s going to fall flat on its face. So get your internal house aligned around why you’re here and what you want to do.”
Implementing “checkpoints” along the way will help ensure stakeholders are still on the same page and that each part of the strategy is agreed upon.
“If you’re bringing your entire organisation along the way and you have a good strategy, then you’ll be able to make the right decision for your organisation as to whether this is the right time to jump in or not,” Hill said.