Meta accelerates the development of new chatbots: focusing on surpassing GPT-4

2023-09-11

According to The Wall Street Journal, Meta has been purchasing AI training chips and building data centers in order to create a more powerful new chatbot, hoping it can be as sophisticated as OpenAI's GPT-4. It is reported that the company plans to start training this new large-scale language model in early 2024, and CEO Mark Zuckerberg is clearly pushing for it to be available for free for company use again, in order to create AI tools. The report points out that Meta has been buying more Nvidia H100 AI training chips and strengthening its infrastructure, so it no longer needs to rely on Microsoft's Azure cloud platform to train new chatbots. The company reportedly formed a team earlier this year to build this model, with the aim of speeding up the creation of AI tools that can simulate human emotions. This goal seems to be a natural extension of Meta's existing generative AI capabilities. Leaked information in June this year revealed that the company is testing an Instagram chatbot with 30 different personalities, which sounds similar to the unannounced AI "personality" that the company is said to be launching this month. According to reports, Meta faced challenges in allocating computing resources among multiple large language model projects this year, leading to a significant turnover of AI researchers. The company faces fierce competition in the generative AI field. OpenAI stated in April that they have not trained GPT-5 and "will not be training for a while," but it is reported that Apple invests millions of dollars every day in developing its own "Ajax" AI model, apparently considering it even more powerful than GPT-4. Google and Microsoft are both expanding the use of generative AI in their productivity tools and hope to incorporate it into Google Assistant. Amazon is also working on generative AI projects internally and may introduce chatbot-driven Alexa.

RECENT AI NEWS

RECENT AI TOOLS