Meta's New AI Model: Aiming for GPT-4's Heights

Image credits: Nick Barclay / The Verge

Meta, the social media giant, is reportedly gearing up to train a new AI model that it hopes will rival the sophistication of OpenAI's GPT-4.

Yes, you heard it right! The tech titan is not just dipping its toes but diving headfirst into the deep end of the AI pool.

According to a recent report by The Wall Street Journal, Meta has been on a shopping spree, snapping up Nvidia H100 AI-training chips like hotcakes. But why, you ask? Well, the company is beefing up its infrastructure to train its new chatbot, aiming to make it as powerful as GPT-4. And this time, it's going solo, without relying on Microsoft's Azure cloud platform.

The plan is to kickstart the training of this new large language model in early 2024. And guess who's cheering them on from the sidelines? None other than CEO Mark Zuckerberg, who is reportedly pushing for the model to be free for companies to create AI tools with.

The goal is to speed up the creation of AI tools that can mimic human expressions, a natural progression of Meta's rumoured generative AI features. Remember the whispers about an Instagram chatbot with 30 personalities? Sounds a lot like the unannounced AI "personas" the company is allegedly launching this month.

However, it's not all smooth sailing. Meta has reportedly faced significant AI researcher turnover due to computing resources being split across multiple LLM projects. Plus, the competition in the generative AI space is fierce. Apple, Google, Microsoft, and Amazon are all vying for the AI crown, investing heavily in their own AI models and initiatives.

So, will Meta's new AI model reach the lofty heights of GPT-4? Only time will tell. But one thing's for sure, the AI race is heating up, and we're all buckled in for the ride!

Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai