The BBC is also planning its own AI model

The BBC is also planning its own AI model
The BBC is also planning its own AI model

Photo by Christoph Meyer (dpa / Scanpix).

The national broadcaster of the United Kingdom, the BBC, plans to create a generative artificial intelligence (AI) model based on its media archive.

The BBC is also negotiating the sale of its archives to large technology companies, which will use the archived material to train a generative AI model.

The Financial Times (FT), which published information about the BBC’s plans last week, claims that the sale of archives for AI model training could be extremely profitable, for example, other media companies Axel Springer and Le Monde made multimillion-dollar deals with the AI ​​company OpenAI for the use of content created by it for generative AI.

The BBC is set to unveil its updated strategy next week, with the UK national broadcaster expected to unveil plans for potential new revenue streams.

According to the FT source, it is likely that the BBC will create its AI model on the basis of an open source, ie publicly available, model.

The Gauss BBC content archives are considered potentially extremely valuable material for training a generative AI model, as the British company’s content is more or less sparse. Some data scientists are afraid of the risks if the models are mainly trained by one or the other side of the political spectrum with more strongly biased source information.

For their part, BBC executives believe that many AI companies are actually already using the British broadcaster’s material without permission, since most of the BBC’s content is available for free.

Back-to-back defenses regarding the media company’s use of content for generative AI model training have been ongoing since 2022. end, when the OpenAI chatbot ChatGPT appears.

Some media companies, such as Politico, Associated Press or Business Insider, agree to use their content for training generative AI models for a fee.

However, in 2023, The New York Times (NYT) by suing OpenAI at the end. The use of content is still in a legal gray area: OpenAI bases its right to use journalistic articles on the valid US fair use a doctrine that allows reasonable use of copyrighted material without consent.

The NYT claims that fair use Not valid in this case, because OpenAI takes large amounts of text.

[{“title”:”Ateitis teism rankose: 5 svarbiausios dirbtinio intelekto bylos”,”url”:”–teismu-rankose-5-svarbiausios-dirbtinio-intelekto-bylos”,”category”:”PREMIUM”,”profiles”:[1019,1040,1067,1633,1644,1660,1686,1873,2213]},{“title”:”Artificial intelligence: not everyone uses it, not everyone will need it, but some risk being uncompetitive”,”url”:” -not-everyone-uses-not-everyone-will-need-but-a-part-risks-being-uncompetitive”,”category”:”PREMIUM”,”profiles”:[1019,1020,1067,1633,1654,1660,1670,1686,1873,2222]},{“title”:”Stanford professor: the decade can be the best or worst in history”,”url”:” this-decade-could-be-the-best-or-worst-in-history”,”category”:”PREMIUM”,”profiles”:[1019,1021,1026,1067,1068,1633,1642,1643,1644,1653,1657,1660,1686,1873]}]

Select the companies and topics you are interested in and we will inform you in a personal newsletter as soon as they are mentioned in Business, Sodra, Registry Center, etc. in altos.

The article is in Lithuanian

Tags: BBC planning model


NEXT As of April 1, the prices of public electricity supply and natural gas are decreasing