Meta Platforms has unveiled the latest iteration of its artificial intelligence models, Llama 3, marking a significant step in the competitive AI landscape. Boasting 405 billion parameters, Llama 3 excels in multilingual conversations, high-quality computer coding, and solving complex math problems, positioning itself as a formidable rival to established players like OpenAI.
While Llama 3 is larger than its predecessor from last year, it remains smaller than some of the leading models, such as OpenAI's GPT-4, which features a trillion parameters, and Amazon's upcoming model with 2 trillion parameters. Despite this, Meta is confident in its AI's capabilities and accessibility.
CEO Mark Zuckerberg announced the development of Llama 4, the successor to Llama 3, which will power Meta's widely used chatbot. Both models will be freely available under an acceptable use policy, allowing other companies to integrate and innovate with Meta's technology. Zuckerberg expressed optimism that future Llama models could surpass proprietary competitors by next year, aiming for Meta's AI assistant to become the most popular by the end of 2024.
In a recent interview with Bloomberg, Zuckerberg emphasized the importance of open and decentralized innovation. He cautioned against isolating the tech sector, highlighting that such measures could hinder progress and the United States' leadership in AI. He also acknowledged the rapid advancements in AI by other nations, noting that maintaining a six to eight-month lead through continuous innovation is a more realistic and attainable goal than trying to stay years ahead.
As Meta continues to push the boundaries of AI technology, the global tech community watches closely to see how Llama 3 and its successors will shape the future of artificial intelligence.
Reference(s):
Meta unveils biggest Llama 3 AI model to rival OpenAI, Google
cgtn.com
