Discussion on Establishing Compensation Systems for Media and Publishing Industries
Amid the surge of generative artificial intelligence (AI) including ChatGPT, media companies and the publishing industry have begun discussing ways to establish a compensation system for the use of news and publications in AI chatbot training.
On the 22nd (local time), The Wall Street Journal (WSJ) reported that the News Media Alliance, representing over 2,000 newspapers in the United States and Canada, has been conducting discussions on this issue over the past several weeks. The meetings focused on how to structure compensation systems and the legal relationships regarding the use of news and publications by AI tools like ChatGPT for training purposes.
Daniel Kopy, Vice President of the News Media Alliance, stated, "We possess valuable content that, through investment, continuously generates revenue for someone. That is truly the work of humans and deserves to be compensated."
Since the launch of ChatGPT in November last year, Microsoft (MS) and Google have successively released AI chatbots. In particular, with MS integrating an AI chatbot into its own search engine Bing, the market where Google had been the 'undisputed leader' is now witnessing a new battle for dominance. For AI chatbots to function effectively as search engines, it appears crucial to rapidly update new information using news.
Individual media companies and news platforms have begun discussions with AI developers. Reddit, a U.S.-based social news aggregation site, is reported to have already started talks with MS regarding news usage fees.
Robert Thomson, CEO of News Corporation, the parent company of WSJ, also revealed earlier this month that discussions are underway with AI companies about monetary compensation for exclusive content usage. Thomson emphasized, "It is a fascinating yet urgent issue. The more AI engines are trained, the more they will utilize content that can make them specialized, and clear compensation must be provided." He explained that the more specialized the AI is in companies and businesses, the more it inevitably relies on Dow Jones content, the WSJ publishing company.
WSJ reported that the core of these discussions lies in whether AI companies’ act of collecting content from the internet to develop AI models is legally permissible. It is important to determine whether the act of AI training itself falls under the U.S. copyright law’s 'fair use,' which allows limited use without the copyright holder’s permission.
Sam Altman, CEO of OpenAI, the developer of ChatGPT, mentioned in an interview that they have given considerable thought to fair use during ChatGPT’s development. He stated that content older than two years was used for training, and if permission was required, contracts related to the content were signed. In fact, ChatGPT has not been updated with information after September 2021, which led to it referring to South Korea’s president as 'Moon Jae-in' earlier this year. Altman said, "For specific domains such as science, we are willing to provide significant compensation for high-quality information."
Regarding Google’s AI chatbot Bard, which was unveiled the day before, when asked to summarize major articles from The New York Times (NYT), it provided a list of some articles and introduced, "For more stories, please visit the NYT website." Sisci Xiao, Vice President in charge of Google Assistant, said, "We are committed to supporting a healthy and vibrant content ecosystem," and added, "We welcome dialogue with stakeholders."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
