Meta plans to match OpenAI’s GPT-4 with its new AI model

Citing GPT-4 as the benchmark, Meta is reportedly gearing up to train a new, highly sophisticated AI model. The company is investing heavily in AI training chips and boosting its data centers to support the ambitious project.

To stay on top of the latest advancements in AI, look here first.

https://preview.redd.it/ts4a6reeuinb1.jpg?width=1440&format=pjpg&auto=webp&s=940be18b0c5f13e2762e1664a96274b314fe00df

Meta’s vision for its new AI model

• Meta's aim is to create a powerful chatbot in line with OpenAI’s GPT-4 capabilities.
• The company has reportedly been in pursuit of Nvidia H100 AI-training chips and is amplifying its infrastructure.
• The idea is to independently train its new model without outsourcing to platforms like Microsoft's Azure.

Efforts and roadblocks

• Meta plans to begin the training of this LLM early in 2024, emphasizing free access for companies.
• Despite its grand vision, Meta has encountered obstacles like researcher attrition and contentious resource allocation amidst multiple LLM projects.
• Noteworthy is the intense competition from major players such as Apple, Google, and Amazon to integrate widely generative AI in their user interface.

Broader implications

• While OpenAI has not immediately revealed plans for a GPT-5, other tech giants are investing heavily. Apple's investment in its "Ajax" AI model signifies the increasing race to advanced AI.
• This move by Meta represents the ongoing trend of tech conglomerates expanding in the AI space, a fact revealed through Google and Microsoft’s use of AI in their productivity tools and Amazon's ongoing developments.

(source)

P.S. If you like this kind of analysis, I write a free newsletter that explores the latest AI developments. Professionals from Google, Meta, and OpenAI are already reading it.