According to JEE News, Meta AI is acquiring training chips and expanding its data centers to develop a more advanced chatbot than OpenAI’s GPT-4.
The tech giant aims to start training new language models as early as 2024, with CEO Mark Zuckerberg advocating free access to AI tools for businesses.
According to the report, Meta is busy expanding its infrastructure and purchasing additional Nvidia H100 AI-training processors so that the new chatbot does not need to be trained on Microsoft’s Azure cloud platform this time.
The corporation assembled a team earlier this year to build models to accelerate the development of AI capabilities that can mimic facial expressions.
According to The Verge, this goal appears to be a logical extension of the creative AI features that Meta is reportedly already developing.
While it’s not certain what Meta will bring to the table as its new AI bot, an Instagram chatbot with 30 personalities was reportedly under testing, according to a June leak.
It sounds a lot like the unknown AI “personality” that businesses are reportedly deploying this month.
According to reports, Meta has experienced a lot of AI researcher turnover this year due to the allocation of computational resources to multiple LLM projects.
Additionally, the field of generative AI is fiercely competitive.
While OpenAI said in April that it had no plans to train GPT-5 anytime soon, it has been reported that Apple is investing millions a day in its AI model, called Called “Ajax,” the company believes it’s even more powerful. GPT-4.
Meanwhile, both Google and Microsoft are incorporating AI into their productivity tools, with Google aiming to use generative AI in its Google Assistant.
Amazon is also working on creative AI initiatives, which could result in an Alexa-powered chatbot.
