Meta's New Generative AI Strategy Puzzles Some While Others Are Excited!

Meta's unconventional AI strategy has raised some eyebrows but some techies rejoice.

Meta's annual Connect conference was held last month when VR fans from all around came to hear Mark Zuckerberg discuss his billion-dollar wager on the metaverse, the technology that will shape Meta's future.

This year's gathering, however, was saturated with panels discussing artificial intelligence instead, a topic that is swiftly moving from the realm of the futuristic to reality.

Joseph Spisak, who had joined the firm as head of product development for generative AI two months before to his session at Connect, remarked, "Don't tell Mark, but it feels less mixed reality and more AI these days." As one attendee put it, "It feels like an AI conference."

The advent of OpenAI's ChatGPT chatbot in November sparked a mad dash by major tech companies to release competing products, including Meta's large language model (LLM) Llama. While Zuckerberg still considers the expansion of the emerging metaverse to be crucial to Facebook's future, artificial intelligence (AI) has become the market he is most focused on today.

Meta considers its Llama series of generative AI tools to be the open-source equivalent of OpenAI's LLM (funded by Microsoft) and Google's PaLM 2 (that drives Bard AI).

Experts in the field view Llama as the potential digital scaffolding supporting the next generation of AI apps, just as Linux software made its way into corporate servers throughout the world and became an integral part of the current internet infrastructure.

According to Twitter post by Yann LeCun, Facebook's new AI research group's head, "This is going to change the landscape of the LLM market."
 

Why Were Some Sceptical Of Meta’s AI Approach?

In contrast to the conventional software licencing and subscription arrangements that made Facebook an internet behemoth, open-source LLMs mean Meta is giving away the software for free to developers.

Meta, in launching Llama 2, stated that a commercial licence would be available for the new version, allowing businesses to incorporate it into their own offerings. Companies such as Microsoft and Amazon, which provide access to Llama 2 as part of their own generative AI corporate offerings, pay the company an undisclosed amount of money despite the company's claims that it isn't focused on monetizing Llama 2.

Mark Zuckerberg has stated that he does not anticipate Llama 2 to create "a large amount of revenue in the near term, but over the long term, hopefully that can be something."

 

How Is Meta Trying To Open-Source Llama?

In a call with analysts in July, Zuckerberg mentioned that third-party development of Llama might lead to "efficiency gains," reducing Meta's operating costs for its AI software. After spending $32 billion in 2018, Meta anticipates much lower capital expenditures of $27 billion to $30 billion in 2023. Susan Li, head of finance, predicted growth in 2024, attributing it in part to spending on data centres and artificial intelligence.

Meta is also placing its faith in an army of volunteer programmers to enhance the performance of Llama 2 and similar artificial intelligence technologies. Ironclad's lead architect Cai GoGwilt has stated that the initial version of Llama was developed by the open-source community to "make it faster and make it run on a mobile phone." The success of Llama 2 depends on the enthusiasm of developers, GoGwilt added.

Meta has also been employing its in-house Research SuperCluster supercomputer to aid in the computationally intensive process of training massive AI models like Llama. The system was created with a whopping 16,000 Nvidia A100 GPUs, the "workhorse" computer chips of the AI business.

A representative from Meta issued the following statement: "We believe in open innovation and do not want to place undue restrictions on how others can use our model." However, we encourage responsible use. This is a custom business licence that strikes a good balance between broad model availability and safeguards against abuse.


As Meta's AI strategy raises eyebrows, will Mark Zuckerberg’s open-source move be the next twist in the AI tale?

Let us know your thoughts in the comments section below!