Image: Meta
Meta has reportedly disbanded its Responsible AI (RAI) team, which was tasked with ensuring the ethical and safe development of artificial intelligence. According to The Information, which cited an internal post, most of the RAI members will join the generative AI product team, which focuses on creating new content using AI. Others will work on Meta’s AI infrastructure, which supports the company’s various AI applications. The company claims that it will still prioritize and invest in responsible AI development and that the former RAI members will continue to support cross-Meta efforts on AI accountability, transparency, safety, privacy, and more.
The RAI team was formed in 2019 to address the potential problems and risks of Meta’s AI training approaches, such as the lack of diversity and fairness in the data and models, and the possible negative impacts of AI on the users and society. However, the team faced several challenges, such as having little autonomy, having to go through lengthy stakeholder negotiations, and facing layoffs earlier this year. As a result, the team was described as “a shell of a team” by Business Insider.
Meta’s decision to break up the RAI team comes at a time when the company’s AI systems have been involved in several controversies and scandals, such as a Facebook translation error that led to a false arrest, a WhatsApp AI sticker generator that produced biased images, and an Instagram algorithm that facilitated the spread of child sexual abuse materials. It also comes at a time when global regulators are trying to establish rules and standards for AI development and use, such as the US government’s AI safety directives and the European Union’s AI principles and AI Act.