However, the decision to open-source Llama has also faced criticism. Some argue that Meta shouldn't use the term "open-source" due to certain restrictions, while others fear that the model could be misused by malicious actors. Despite these concerns, Llama has proven successful, with more than 90% of open-source AI models currently in use being based on Llama. The company is now working on Llama 4 and has plans to develop artificial general intelligence (AGI).
Key takeaways:
- Meta's generative AI model, Llama 2, was released as an open-source model in the summer of 2023, despite concerns about potential misuse and the challenge of monetizing a free product.
- The decision to open-source Llama 2 was made by CEO Mark Zuckerberg, who believed it would enable the model to become more powerful more quickly and at a lower cost, helping Meta catch up in the generative AI race.
- By the time Llama 3 models were released in 2024, Llama had mostly caught up to its closed-source rivals in speed and accuracy, and was being used to power many of Meta's products.
- Despite the high costs of developing Llama, Meta is giving it away for free to thousands of companies, including giants like Goldman Sachs, AT&T, and Accenture, with the belief that it will attract top talent, accelerate innovation, develop new revenue sources, and extend its longevity.