According to the Financial Times, as Meta CEO Mark Zuckerberg invests more money in artificial intelligence development, the company is working to replace human content moderators with large language models, a move that is part of its overall plan to reduce operating costs.
Meta
The $1.4 trillion technology company has accelerated plans to use large language models to moderate content and ads on its platform, according to four people familiar with the matter. The shift is expected to save the company billions of dollars annually.
Meta has replaced about 50% of manual review requests with large language models this year, and the company plans to further reduce human involvement by the end of the year, people familiar with the matter said. For some types of content, manual review rates may be reduced by more than 90%.
Meta has long used a combination of automated systems and human reviewers, including third-party contractors, to determine whether posts or ads violate its platform rules. Appeals raised by users are usually handled by human reviewers.
As Meta advances its AI content moderation plan, Zuckerberg is investing billions of dollars in talent recruitment and infrastructure construction to develop what he calls "personal superintelligence," a technology that promises to provide users with highly personalized AI products and agents.