Meta is developing a new, open-source AI model aimed at outperforming OpenAI’s advanced systems by 2024.
Meta is actively constructing a groundbreaking, open-source artificial intelligence (AI) model to compete with OpenAI’s most potent systems. Sources reveal that this yet-to-be-named project will significantly surpass Meta’s previous AI model, Llama 2, in terms of power and capabilities. With plans to release this cutting-edge technology sometime next year, Meta is setting up the infrastructure to make it happen, including stocking up on Nvidia’s top-tier H100 semiconductor chips.
The Wall Street Journal recently broke the news, citing that people with knowledge of the matter indicated Meta’s new model would have multiple times the power of Llama 2. Released earlier this year, Llama 2 was trained on 70 billion parameters. In contrast, OpenAI’s latest release, GPT-4, reportedly operates on an estimated 1.5 trillion parameters. The drive to compete with OpenAI and other major players in the AI arena is clear, and Meta aims to commence the training phase for this AI marvel by the beginning of 2024.
Open Source for a Collaborative Future
Meta’s decision to make the new AI model open-source is noteworthy. The company aims to offer other businesses the ability to construct their AI tools for high-quality text generation, data analysis, and more. This approach could disrupt the AI industry, encouraging a wave of innovation and cooperative development.
To achieve these lofty goals, Meta is not cutting corners. The company is expanding its data centers to handle the computational load this new model will require. They are also buying more of Nvidia’s H100 semiconductor chips, which are the most powerful chips currently in the market. The company plans to use its infrastructure for training, indicating a shift from its previous collaboration with Microsoft’s Azure platform for Llama 2.
The Global Race for AI Supremacy
This announcement comes at a time when both governments and corporations are pushing the boundaries of what is possible with AI technology. The United Kingdom recently declared an investment of $130 million in high-powered chips for AI systems. Meanwhile, China has enacted new AI legislation, leading to the release of over 70 AI models in the country, according to the CEO of Baidu.
As the competition intensifies, Meta’s ambitious project places it squarely in the race for AI dominance. With plans to outdo even OpenAI, the tech world will undoubtedly be watching closely to see what unfolds.