BBC plans to build its own AI model and negotiate data access rights with tech giants such as Amazon

Wallstreetcn
2024.03.22 06:28
portai
I'm PortAI, I can summarize articles.

Once the agreement to sell database access rights is reached, it will bring substantial revenue to the BBC

AI is accelerating its penetration into the global media industry.

According to sources cited by the media, the BBC (British Broadcasting Corporation) is in talks with tech giants including Amazon to sell database access rights, discussing the possibility of using its content for AI training.

Once an agreement is reached, it will bring substantial revenue to the BBC. Some believe that due to the relatively neutral nature of BBC's news content over the years, using it as a data resource for training AI models is particularly valuable.

The reason behind BBC's decision may be that since most of the content is freely available, many AI companies may already be using BBC's content to train models.

Previously, media groups such as Axel Springer and Le Monde have reached multimillion-dollar deals with OpenAI to open up news content for training AI models, and Reuters has also signed multiple similar agreements with AI developers.

Meanwhile, BBC is also planning to develop its own AI models. Rhodri Davies, BBC's Director of National Operations, stated earlier this month that they are evaluating the possibility of developing AI technology through collaboration or independently. Davies also mentioned that cost is an important consideration:

"We also need to evaluate the costs. Clearly, the BBC has nearly 80 years of audiovisual archives and a vast amount of online text, and we are working to assess the feasibility of relevant solutions."

It is reported that BBC is testing the use of AI in editorial rooms internally (such as automatically generating news headlines), but any content must undergo manual review before being published.

Currently, the BBC has stated that it has not reached any formal agreements with any organization:

"We have not reached any agreements with any organization to use our data for training commercial generative AI models."

"We are evaluating the potential bias issues in such models and how the BBC should address them in the future. At this point, we cannot determine whether to choose collaboration or independent solutions."

Additionally, sources have indicated that if the BBC chooses to develop AI projects through its own research and development team, it may leverage currently available "open-source" AI models