Following Mark Zuckerberg’s announcement on Instagram regarding Meta’s new Meta AI assistant, powered by Llama 3, experts have raised concerns over the significant increase in energy consumption required to develop the new model. Meta revealed that Llama 3 was fed seven times as much data as its predecessor, Llama 2, in an effort to improve the training data. However, the company’s release of figures showing the massive energy consumption for creating Llama 3 has drawn criticism from AI experts.
The advancement of open source AI models, such as Llama 3, has sparked worries among experts about the potential risks associated with their growing capabilities. Some experts have expressed concerns that these models could be used to develop cyber, chemical, or biological weapons, or even pose a threat to humans. In response to these concerns, Meta has released tools to ensure that Llama does not output harmful utterances. However, some AI researchers argue that Meta’s Llama models are not as open as they could be due to restrictions placed on the models’ open source license.
Stella Biderman, an AI researcher involved with EleutherAI, has criticized Meta’s licensing terms for Llama 2 and Llama 3, stating that the restrictions limit the experiments that AI researchers can conduct with the models. Biderman points out that Meta’s licensing terms, particularly for Llama 3, are more restrictive, with clauses that hinder experiments for companies with “greater than 700 million monthly active users.” This restriction is seen as an effort to prevent the models from benefiting Meta’s competitors. Despite these limitations, the release of Llama 3 is expected to stimulate a new wave of AI experimentation.
Anticipated Impact on AI Development
Clement Delange, CEO of HuggingFace, a repository for open AI models, including Llama 3, predicts that the release of Llama 3 will lead to the development of numerous new models based on its technology. Delange highlights that developers created over 30,000 variants of Llama 2 and expects a similar level of innovation with Llama 3. While the restrictions in Meta’s licensing terms have raised concerns among AI researchers, the potential for new advancements in AI development remains a strong possibility.
Meta’s introduction of Llama 3 as part of its new Meta AI assistant has stirred both excitement and apprehension within the AI community. The advancements in training data and capabilities of the model have been met with concerns over energy consumption, restrictions in the open source license, and limitations for AI researchers. Despite these challenges, the release of Llama 3 is expected to drive innovation and experimentation in the field of AI, paving the way for new developments and advancements in the future.
Leave a Reply