Meta Launches Open Innovation AI Research Community with Limited Resources
Tech giant Meta is stepping into the AI field with the launch of the Open Innovation AI Research Community. The community aims to foster transparency, innovation, and collaboration among AI researchers. Initially, the focus will be on the privacy, security, and safety of large language models such as OpenAI’s ChatGPT. Additionally, the group will provide input into AI model refinement and set an agenda for future research.
Meta plans to have its own researchers participate in the community, but it will be member-led with Meta AI acting as a facilitator. The community will champion large open-source foundation models and provide a platform for partners to collaborate and share learnings. It also aims to accelerate the training of the next generation of researchers.
While Meta intends to sponsor workshops and develop guidelines for responsible open-source model development, the details beyond that remain vague. There is a possibility of creating a website, social channels for collaboration, and research submissions to academic conferences, although no commitments have been made.
The Open Innovation AI Research Community expects its members to fund their own work. Meta has not indicated that it will allocate resources or computing power specifically for the community. This approach may be an attempt to avoid potential bias or undue influence. However, considering the high expenses associated with AI research, funding may be a significant obstacle for participants.
Given Meta’s recent controversies and criticisms surrounding AI, the establishment of the Open Innovation AI Research Community raises questions about its sincerity and level of commitment. Meta faced backlash last year when it had to remove an AI demo that generated racist and inaccurate content. Its AI ethics team has been criticized for being ineffective, and the anti-bias tools released by the company have been deemed completely insufficient. Academic circles have accused Meta of exacerbating socioeconomic inequalities with its ad-serving algorithms and displaying bias against Black users in automated moderation systems.
It remains to be seen whether the Open Innovation AI Research Community will address these concerns. Meta is encouraging professors at accredited universities with relevant AI experience to participate. However, there are already numerous open machine learning research communities unaffiliated with big tech companies, which may offer more attractive options for researchers.
The deadline for applying to the Open Innovation AI Research Community is September 10. Meta welcomes applicants from diverse research disciplines and those with technical capabilities for research. Multiple participants from the same university are allowed to apply. It is yet to be determined whether Meta’s efforts will truly deliver on its promises to create robust and representative models in the AI field.