The Ministry of Digital Development and Development has promoted the "Taiwan Sovereign AI Training Corpus" private corpus collection plan. Legislators are concerned about issues such as related authorization, payment mechanisms and media rights protection if news content is included in the AI training corpus in the future. (Photo by reporter Qiu Qiaozhen, file photo)
[Reporter Qiu Qiaozhen/Reporting from Taipei] The Ministry of Digital Development has previously held a private corpus collection activity for the "Taiwan Sovereign AI Training Corpus" to strengthen the local cultural connotation and corpus depth of the Taiwan Sovereign AI Training Corpus and promote the private corpus collection plan. However, the relevant corpus authorization and payment mechanisms have also caused concern. Legislators raised concerns during the inquiry today that if the scope of the collection is expanded to include news media in the future, the industry may be required to provide content for free, which will affect the rights and interests of the news industry.
Legislator Lin Chuyin said during the Transportation Committee inquiry that she was born in the media and is well aware that electronic and print media are currently affected by changes in the online advertising market. The Ministry of Data Development is currently collecting corpus from the publishing industry and e-book platforms on a voluntary basis. Will it further expand to electronic and print media in the future, so that news content can become free AI training materials? Therefore, it is hoped that the Ministry of Data Development will explain the future planning direction of "data has a price".
Please read on...
In this regard, Lin Yijing, Minister of the Department of Digital Development, emphasized that what is currently being promoted is "recruitment" rather than "recruitment", and full voluntary participation is adopted. The folk corpus collected at this stage is mainly divided into four categories, including: public books whose copyright protection period has expired, book introductions provided by publishers, some book trial contents, and works voluntarily authorized by individual authors.
Lin Yijing explained that the Digital Development Department obtained authorization for AI training and did not require authors to donate copyrights. The two concepts are different. As for news media content, it has not yet been included in the scope of fundraising. From his understanding, AI companies are usually more willing to pay to obtain media content. International media have already signed paid licensing agreements with large AI companies. The Ministry of Digital Development will try its best to promote relevant cooperation in the future.
Legislators are further concerned that when an author or publisher provides corpus for AI training, and if the training results are derived into other products, how should the rights and interests of the original content provider be protected to avoid the situation of users not paying. In addition, if authorized works choose to withdraw in the future, how to protect the rights and interests of both parties, and how to establish a reasonable payment mechanism for news content, all should be planned in advance.
Lin Yijing responded that the ultimate goal of the Ministry of Digital Development is to hope that AI companies, especially large American AI companies, can pay to obtain relevant training content. However, when emerging technologies impact the property rights of works, relevant disputes often last three to five years. However, Taiwan’s AI development cannot wait. Therefore, at this stage, we hope to prioritize the non-controversial parts and avoid involving controversial content. This is also the reason why the Ministry of Digital Development emphasizes voluntary participation. Currently, it mainly targets less controversial corpus such as public books and book introductions.
The legislator finally stated that relevant media and communication units under the Ministry of Culture also have corpus that can be used, and the possibility of cooperation can be further discussed in the future. Although the construction of AI corpora must be gradual, the government should still explain the licensing and payment principles in advance to reduce the doubts of the publishing and news industries and avoid disputes from becoming an obstacle to the subsequent promotion of sovereign AI and corpus construction.
Grasp the economic pulse with one hand. Click here to subscribe to Free Finance Youtube Channel
