Meta dismisses RAI unit that was looking for problems in AI training approaches

Meta has disbanded its Responsible AI (RAI) division. He was engaged in identifying problems in approaches to AI training. This was reported by The Verge with reference to The Information.

Some of RAI’s employees will now move to the generative artificial intelligence division, while others will work on Meta’s AI infrastructure.

Despite the reorganization, Meta promises to continue paying attention to the responsible development of AI. According to the company’s representative John Carville, Meta “will continue to prioritize and invest in the safe and responsible development of AI.”

As you know, the RAI team has been working since 2019. It was concerned with whether the company’s models were trained on adequately diverse information. This was done to prevent problems with moderation on Meta.

But, according to the publication, the team of this unit had limited autonomy. Therefore, long negotiations with stakeholders had to take place before the proposed initiatives could be implemented.