AI Model Turned Military Tool in China: A Surprising Twist
ChinaSun Nov 03 2024
Chinese military researchers have found a new use for Meta's Llama AI model. They've adapted it to create a tool called "ChatBIT, " which is designed to help with military tasks like gathering intelligence and making decisions. This is a big deal because Meta shared Llama publicly to encourage open innovation, but they didn't expect it to be used for military purposes.
The Chinese researchers are from institutions linked to the People’s Liberation Army (PLA). They've been working on ChatBIT since June, using an early version of Llama as its base. They say it's good at answering questions in a military context, but they haven't revealed how it's actually being used.
Sunny Cheung, an expert from the Jamestown Foundation, thinks this shows a shift. PLA scientists are now systematically adapting open-source AI models for military use. Meta has said that this isn't allowed and goes against their policies. They argue that while their models promote transparency, China's AI investments make any single model less valuable.
ChatBIT currently relies on a dataset of 100, 000 military dialogues, which is smaller than what other large-scale models use. Joelle Pineau, Meta's VP of AI Research, questioned ChatBIT's capabilities due to this smaller dataset. She pointed out that vast amounts of data are crucial for improving model reliability.
China's advancements in AI, especially in military and security areas, have raised concerns in the US. President Biden recently signed an executive order to regulate AI developments. The US is also proposing rules to control investment in Chinese AI and other critical sectors.
The Pentagon sees both opportunities and risks in open-source models. They're keeping a close eye on what other countries are doing. William Hannas of Georgetown University's Center for Security and Emerging Technology thinks that as China speeds up its AI research, it will be harder to restrict access due to ongoing collaboration between Chinese and Western scientists.
https://localnews.ai/article/ai-model-turned-military-tool-in-china-a-surprising-twist-fc62a061
continue reading...
questions
What steps can be taken to ensure that open-source models are used responsibly and within intended guidelines?
Could Meta's open-source models be a secret way to infiltrate foreign militaries?
If ChatBIT becomes self-aware, will it ask for a military discount on toothpaste?
actions
flag content