The BBC, Britain’s national broadcaster, is exploring the possibility of constructing and training its own artificial intelligence model utilizing its vast text archives.
Artificial intelligence, machine learning – artistic interpretation. Image credit: Mahdis Mousavi via Unsplash, free license
The corporation revealed yesterday the organization’s thoughts on employing generative AI as a supportive tool in its production process. Official representatives also mentioned evaluating potential arrangements, either independently or through partnerships, for training such a model.
Responding to a report by the Financial Times on Thursday, which suggested that the BBC was considering developing its own AI models, a spokesperson confirmed the corporation’s interest in creating a Large Language Model based on its text-based content. Moreover, the spokesperson clarified that any output generated by such a model would be exclusively accessible to the BBC.
Also, discussions are underway between the broadcaster and technology firms regarding the potential sale of access to its archives. These archives could then be utilized by the firms to train AI models. In addressing this report, the BBC emphasized that it has not entered into any agreements yet permitting organizations to utilize its archive for training large language models intended for commercial purposes.
The BBC underscored its commitment to thoroughly examining various aspects associated with such initiatives, including potential biases within AI models. They aim to explore how the BBC, whether independently or in collaboration with others, can effectively address these considerations.