Report: Chinese researchers used public Meta AI to build model for military purposes
Chinese researchers leverage Meta AI for military AI model
According to Reuters, Chinese researchers associated with the People’s Liberation Army (PLA) have utilized Meta's large language model Llama for developing an AI tool with potential military applications. The researchers utilized an early version of the Llama model, known as Llama 2 13B, to create "ChatBIT" - a military-focused AI tool optimized for dialogue and question-answering tasks in the military domain.
ChatBIT's performance and capabilities
The ChatBIT model, as discussed in a research paper by six Chinese researchers, displayed high performance in comparison to other AI models. While the exact capabilities of ChatBIT remain undisclosed, the researchers highlighted that the model was trained on only 100,000 military dialogue records, a significantly lower amount than other large language models available.
Concerns and responses
Joelle Pineau, Meta's Vice President of AI Research, expressed skepticism about the model's capabilities due to its limited training data compared to other models in the field. Additionally, Meta has enforced restrictions on the use of its AI models, particularly in sensitive areas such as military and espionage applications.
Despite global competition in the AI sector, Chinese investments in AI technology continue to grow, with the alleged use of an outdated American open-source model deemed irrelevant in this context.
Meta's stance on model use
While Meta promotes the accessibility of its AI models, the company has imposed limitations on their usage, requiring licensing for services with a large user base. Meta's director of public policy emphasized that any unauthorized use of their models, especially by the People's Liberation Army, contradicts the company's acceptable use policy.
Source: Reuters
