M
eta is reportedly making significant strides in the development of an advanced AI language model aimed at rivaling OpenAI’s ChatGPT. According to sources cited by The Wall Street Journal, this new AI tool will be designed to generate text and analyses with a focus on serving businesses. Meta’s move is seen as an attempt to stay competitive in the ever-evolving field of AI, which is witnessing the emergence of numerous new models and technologies.
The primary goal of Meta’s forthcoming AI model is to provide businesses with a suite of services, including the generation of sophisticated textual content and in-depth data analysis. To bolster their AI capabilities, Meta is said to be working on acquiring H100s, advanced Nvidia chips known for their prowess in training AI software. These chips will play a crucial role in Meta’s efforts to build a cutting-edge AI model that can compete with the likes of ChatGPT.
This development follows Meta CEO Mark Zuckerberg’s previous partnership with Microsoft, which owns OpenAI. Their collaboration resulted in the creation of Llama 2, an AI language model built using Microsoft’s Azure platform. Shortly after the release of Llama 2, Meta has embarked on the journey to develop a revamped AI product that is expected to be unveiled next year.
Despite facing stiff competition from other AI giants like OpenAI’s ChatGPT and Google’s Bard, Meta is reportedly planning to make its AI tool open source. This decision aligns with Meta’s vision of providing free access to users, allowing businesses, startups, entrepreneurs, and researchers to harness the power of advanced AI tools without cost barriers.
During a July earnings call, Mark Zuckerberg expressed optimism about Meta’s “groundbreaking AI products in the pipeline.” He emphasized the importance of AI in Meta’s strategic focus and the significant investments they have made in AI infrastructure. These investments have yielded improvements in ranking and recommendation systems, ultimately enhancing user engagement and monetization.
However, as companies like Meta race to develop increasingly advanced AI systems, concerns about the reliability of these tools are growing. Recent reports indicate that AI models, including GPT-4, are experiencing declines in accuracy. For example, a study by researchers at Stanford University and UC Berkeley revealed that GPT-4’s accuracy in identifying prime numbers dropped significantly from 97.6% in March to just 2.4% in June.
Meta’s response to these concerns and the evolving AI landscape will be closely watched as they continue to develop their advanced AI language model. With plans to offer it as an open-source tool, Meta aims to democratize access to AI capabilities, potentially opening up new opportunities for businesses and researchers alike. As they move forward, Meta is determined to build a strong foundation for the next generation of AI products, reaffirming their commitment to the rapidly advancing field of artificial intelligence.
