The BBC, the national broadcaster of Britain, announced on Thursday that it is developing plans to use its text archives to create and train an artificial intelligence model.
Earlier this month, a BBC executive testified before a parliamentary committee that the company was considering generative AI as a tool to enhance its production process and was considering a potential unilateral or cooperative model-training agreement.
When questioned about a Financial Times story from Thursday claiming the BBC intended to create its own AI models, a representative for the organization acknowledged the organization was considering creating a Large Language Model with text-based material.
The representative stated that only the BBC would have access to the output of any such model.
The broadcaster was in talks with tech companies to sell access to its archives, which might be used to train AI models, according to a story by The Financial Times that cited sources.
The BBC responded by saying that it has no agreements with any organizations to train its huge language models, which are the basis for generative AI products intended for commercial use, using its archive.
“We are looking at a wide range of issues such as potential bias within such models and how the BBC either in partnership or unilaterally can address these.”
