Meta and IBM have joined forces, launching the ‘AI Alliance’ alongside over 50 other entities, including tech giants, universities, and organizations. This alliance aims to revolutionize the development of AI models, making them more secure, open, and accessible.
The AI Alliance is supported by industry heavyweights like Oracle, Intel, AMD, the Linux Foundation, and even NASA. However, the notable absence of NVIDIA, a leader in AI data center hardware, raises eyebrows.
This alliance aims to unite efforts in developing resources like tools, benchmarks, and evaluation standards for global application in AI systems development and usage, with a particular emphasis on security. The members of the AI Alliance are not just looking to create foundational models free from proprietary technology limitations. They aim to share these models, allowing researchers to build upon each other’s work, thereby accelerating progress. This collaborative approach also seeks to simplify the process of identifying and rectifying code flaws or vulnerabilities.
One of the most compelling aspects of this alliance is its potential to create a level playing field in AI development. The open-source approach offers an equitable starting point for all players in the sector, regardless of their size or resources. While this is an attractive proposition in theory, its practical application remains to be seen.
The AI Alliance is committed to fostering an open community, aiming to accelerate responsible AI innovation while ensuring scientific rigor, trust, security, diversity, and economic competitiveness. By bringing together top developers, scientists, academic institutions, companies, and other innovators, the alliance plans to pool resources and knowledge to address security concerns and provide a platform for sharing and developing solutions that cater to the needs of researchers, developers, and adopters globally.
Meta’s involvement in this alliance is particularly significant, considering its push for open-source AI in 2023. The company has made notable strides with releases like LLaMA 2, a model that competes with GPT-4, and tools such as AudioCraft and Code Llama. However, Meta’s efforts have been met with some controversy. The Open Source Initiative criticized the licenses for LLaMA 2 and related AI developments for not being genuinely open-source, as Meta restricted their use for certain purposes, particularly commercial ones.
Despite these challenges, the AI Alliance has been well-received by its founding members. David Spergel, president of the Simons Foundation, highlighted the critical role of open-source AI in advancing science and enabling ethical use. Jeremy Howard, a founding researcher at Fast.ai, pointed out that open-source is the “backbone of all leading AI.”