Meta is about to destroy OpenAI’s business model by making its incoming, open-source LLM completely free. Here’s everything you need to know…
KEY TAKEAWAYS
- Meta is set to issue a commercial license for its upcoming open-source large language model (LLM).
- This move could disrupt the AI landscape, currently dominated by closed-source models like Google’s Bard and OpenAI’s ChatGPT.
- Meta’s previous open-source LLM, LLaMA, has already inspired a new generation of AI models.
- Open-source models are rapidly catching up with their closed-source counterparts.
- OpenAI is reportedly preparing its own open-source LLM in response to the competition.
In a move that could potentially reshape the AI industry, Meta is said to be on the brink of monetizing its forthcoming open-source language model. This development, first brought to light by The Information, indicates a possible paradigm shift in an AI field largely controlled by proprietary models such as Google’s Bard and OpenAI’s ChatGPT.
Meta’s intention to grant a commercial license for its soon-to-launch, open-source Large Language Model (LLM) could lay the groundwork for a more cost-effective and adaptable AI solution. This initiative is anticipated to draw in businesses looking to harness the power of AI without the substantial financial burden that comes with proprietary models.
Open Source LLMs vs Closed Source LLMs (Like Bard and ChatGPT)
The advantages of Meta’s approach are not limited to the businesses that adopt open-source models. As an increasing number of developers immerse themselves in Meta’s open-source AI environment, the company is poised to benefit from the collective knowledge of AI engineers globally.
Think about how Linux and other open-source platforms like WordPress work. Now, imagine applying that model to scalable AI? It’d speed up development, boost productivity, and fast-track myriad new uses and capabilities for users – it seems like a massive win-win.
Building On From LLaMA
Meta’s progression towards monetizing open-source AI is not entirely unexpected. The company has a track record of experimenting with open-source AI, with its LLaMA model, launched in February 2023, serving as a notable example.
Initially licensed for research purposes and shared with a limited user base, LLaMA’s code eventually permeated the wider tech sphere, sparking a new wave of open-source AI models. And with Meta’s new ChatGPT-like LLM coming, the writing could well be on the wall for Google and OpenAI’s closed approach to generative AI.
Are Open-Source Models The Future?
Open-source models are swiftly narrowing the gap with their proprietary counterparts. Take, for example, the Vicuna LLM model, constructed on the LLaMA foundational model, which purported to deliver 90% of ChatGPT’s quality in March.
Even one of Google’s AI engineers – a chap called Luke Sernau – noted, inside a confidential internal memo, that the edge held by Google and OpenAI with their proprietary models was eroding.
The open-source AI scene has witnessed even more advancements since then. Hugging Face’s Open LLM leaderboard now showcases a variety of advanced models, many of which are offshoots of the original LLaMA LLM. These models persist in enhancing their capabilities at a remarkable pace, with models like Falcon, developed by Abu Dhabi, gaining traction among developers.
In response to the escalating competition from open-source models, OpenAI is reportedly gearing up to launch its own open-source LLM. This action is perceived as an effort to safeguard its market dominance in the face of mounting pressure from open-source rivals.
Open Source vs Closed Source AI Models: Which is Best?
Before we delve into the pros and cons, let’s first understand what open-source and closed-source AI models are.
Open-Source AI Models
Open-source AI models are those whose source code is publicly accessible and can be freely used, modified, and distributed by anyone. These models are often developed by a community of programmers who contribute to the codebase, making improvements and adding new features.
Examples of open-source AI models include TensorFlow (developed by Google), PyTorch (developed by Facebook), and Hugging Face’s Transformers, which provide pre-trained models for Natural Language Processing (NLP) tasks.
Closed-Source AI Models
Closed-source AI models, on the other hand, are proprietary models developed by specific companies or individuals. The source code for these models is not publicly available, and usage rights are typically granted under a license. These models are often commercialized, and users may need to pay to use them or access their full functionality.
Examples of closed-source AI models include Google’s BARD, OpenAI’s GPT-3, and IBM’s Watson.
Pros and Cons of Open-Source AI Models
Pros
- Accessibility and Affordability: Open-source AI models are freely available, making them accessible to anyone, from individual developers to small startups. This democratizes AI development, allowing even those with limited resources to build AI applications.
- Community Support: Open-source models benefit from a large community of developers who contribute to the codebase. This means that bugs are often quickly identified and fixed, and new features are regularly added.
- Transparency: With open-source models, users can inspect the code to understand exactly how the model works. This transparency can be crucial for applications where explainability is important.
Cons
- Quality and Performance: While open-source models can be highly effective, they may not always match the performance of proprietary models developed by tech giants with vast resources.
- Support and Maintenance: While the community can provide support, there’s no dedicated customer service in case of issues or queries. Also, maintaining and updating the model depends on the community, which can be inconsistent.
- Security: Open-source models can be more vulnerable to malicious attacks since the code is publicly accessible.
Pros and Cons of Closed-Source AI Models
Pros
- Performance: Closed-source models, especially those developed by large tech companies, often deliver high performance. They are developed by dedicated teams with substantial resources, leading to advanced features and capabilities.
- Support: Users of closed-source models often have access to dedicated customer support from the company, which can be beneficial in resolving issues or answering queries.
- Security: The source code of closed-source models is not publicly accessible, which can make them less vulnerable to attacks.
Cons
- Cost: Using closed-source models can be expensive, especially for small businesses or individual developers. Some models require a subscription or usage fees.
- Lack of Transparency: With closed-source models, the underlying code is not visible to users. This lack of transparency can make it difficult to understand how the model is making decisions.
- Dependency: Users of closed-source models are dependent on the provider for updates and improvements. If the provider decides to stopsupporting the model, users may be left with an outdated tool.
Meta: The Unlikely Pioneer of Open-Source AI?
As AI continues to play a crucial role in Meta’s strategic blueprint, the company is reinforcing its commitment to an open-source approach.
This commitment is manifested in Meta’s consistent release of research and code to the public, a stark contrast to Google and OpenAI’s strategy of secluding much of their research.
🤖 Curious about AI? Stay on top of ALL the latest AI news, tools, and trends with our tri-weekly newsletter – it goes out every Monday, Wednesday, and Friday.
