A lot of folks are wondering why it is that Meta (formerly Facebook) is releasing open-source Large Language Models (LLMs) for free. In reality, Meta's decision to release these LLMs aligns well with their history and business strategy. Here are three key reasons:
History of Open Source: First, Meta has a solid history of open-sourcing very cool tech. Look no further than their extensive list of contributions. I dare you to click "view all" and scroll to the bottom. Go on. I'll wait. The list is long and the point is that points to a history of open-sourcing quite a bit of technology. Previous major examples include React and PyTorch. The Llama models follow this consistent support.
Enhancing User Engagement: Meta makes money as users spend more time on their platforms. Generative AI tools, like the LLMs and image generators, increase user interaction and content sharing across social networks, which directly boosts Meta's business. This is also why we see them releasing new experiences like characters. These are experiments to find new ways of engaging their users.
Reputation and Branding Strategy: Following several privacy and data scandals, Meta needs to rebuild its reputation. By openly supporting open-source AI, they position themselves as a leader in the field, contrasting with closed-source alternatives. Mark Zuckerberg, in particular, seeks to enhance his and his companies' recognition by being at the forefront of bringing open-source AGI to the public. If he can shift public opinion by being the one who brings open-source AGI to the masses, well, that may be simply too tempting a narrative to pass up.