Meta Platforms Inc. executives have stated that sharing their AI technology openly has not posed any significant disadvantages, in contrast to many of their peers. Meta has been releasing open-source versions of its large language models over the past few months, allowing others to build products and services on top of them. This approach sets Meta apart from other major technology companies like OpenAI and Google’s DeepMind, which currently do not open-source their large language models.
While some companies fear that sharing their work could result in competitors stealing it, Meta believes that staying open offers advantages. By making their AI models accessible to thousands of developers worldwide, Meta can tap into a wider pool of expertise and accelerate the improvement of their technology. Yann LeCun, Meta’s chief AI scientist, emphasized the importance of transparency, stating, If you do this in secret, you fall behind.
Meta’s decision to provide free access to its large language models, which require substantial computing power, allows smaller companies and developers to build AI tools. Although Meta plans to charge big companies like Microsoft, Amazon, and Google for using their models, they do not expect this to generate significant revenue in the near term. Meta CEO Mark Zuckerberg expressed hope that it could become a source of income in the long run.
Meta’s AI research Lab FAIR, or Facebook Artificial Intelligence Research, established over a decade ago, now boasts over 400 researchers across North America and Europe. Recently, the team has been focusing on finding practical applications for their research, resulting in the creation of a new AI product group dedicated to integrating generative AI technologies into Meta’s products.
By openly sharing its AI technology, Meta is not only advancing the field but also fostering collaboration and innovation among developers. The approach aligns with Meta’s vision of bringing people closer together through technology while keeping pace with rapid advancements in the AI landscape.
