Meta sets GPT-4 as the bar for its next AI model, says a new report

Next AI Model: Meta Sets Gpt-4 as the Bar for Its | Enterprise Wired

Share Post:

LinkedIn
Twitter
Facebook
Reddit
Pinterest

According to The Wall Street Journal, Meta has been actively acquiring AI model training chips and expanding its data center infrastructure with the aim of developing a more advanced chatbot. The company hopes to create a chatbot that can rival the sophistication of OpenAI’s GPT-4. Their plan is to commence the training of this new large language model in early 2024. CEO Mark Zuckerberg is reportedly advocating for it to once again be available for companies to use in the development of AI tools without charge.

Related to the Instagram chatbot with 30 Personalities

The Journal reports that Meta has been increasing its procurement of Nvidia H100 AI-training chips and strengthening its infrastructure to reduce reliance on Microsoft’s Azure cloud platform for the chatbot’s training. Earlier this year, Meta assembled a team dedicated to building this model, with the primary objective of accelerating the development of AI tools capable of mimicking human expressions. This aligns with Meta’s ongoing efforts in generative AI, as evidenced by rumors of “personas” it plans to launch later this month, which may be related to the Instagram chatbot with 30 personalities that were leaked in June.

Meta has reportedly faced challenges related to AI researcher turnover due to resource allocation among multiple large language model projects this year. Additionally, Meta faces fierce competition in the generative AI field. While OpenAI announced it wasn’t currently training a GPT-5 in April, Apple is investing heavily in its “Ajax” AI model, which it believes surpasses even GPT-4 in terms of power. Google and Microsoft are also expanding their use of AI in productivity tools, with Google aiming to incorporate generative AI into Google Assistant. Furthermore, Amazon is actively pursuing generative AI initiatives throughout its organization, potentially leading to a chatbot-powered Alexa.

AI model ready for deployment next year

Meta Platforms is reportedly in the process of developing a new artificial intelligence system with aspirations to match the capabilities of OpenAI’s most advanced model, according to insiders cited by the Wall Street Journal.

The social media giant’s goal is to have its upcoming AI model ready for deployment next year, and it is anticipated to be significantly more potent than its current commercial iteration, known as Llama 2. Llama 2 was introduced by Meta in July and is distributed through Microsoft’s Azure cloud services, positioning it in competition with OpenAI‘s ChatGPT and Google’s Bard.

While the specifics of this planned system may evolve, it is intended to facilitate the development of advanced text generation, analysis, and other services by external companies, as reported by the newspaper.

The training of this new AI system, categorized as a large language model, is expected to commence early in 2024, as per the report.

RELATED ARTICLES

Ford to Cut 800 Jobs in the UK Amidst Weak EV Demand

Ford to Cut 800 Jobs in the UK Amidst Weak EV Demand

Restructuring Across Europe Ford has announced plans to cut 800 jobs in the UK over the next three years as…
Content Creation for Business Owners: A No-Nonsense Guide to Success

Content Creation for Business Owners: A No-Nonsense Guide to Success

You’re staring at the blinking cursor again. A blog post is due, the email newsletter needs writing, and the social…
Nifty and Sensex Rebound, Signal Stabilization in Indian Markets

Nifty and Sensex Rebound, Signal Stabilization in Indian Markets

Positive Opening for Indian Indices After a streak of losses, Indian stock markets opened on a positive note on Tuesday,…
Bluesky’s Rapid Growth: A New Chapter in Social Media Evolution

Bluesky’s Rapid Growth: A New Chapter in Social Media Evolution

A Decentralized Vision Takes Shape Bluesky’s, a new player in the social media landscape, is making waves as users look…