Chinese tech giant Baidu has denied any connection with a Chinese military laboratory allegedly using commercial large language models (LLMs) for advancing artificial intelligence (AI) in military applications.
The denial comes in response to media reports, first published in the South China Morning Post, surrounding an academic paper that mentioned Baidu’s ERNIE Bot in connection with the People’s Liberation Army’s (PLA) Strategic Support Force.
In the paper, scholars from a Chinese university detailed how they employed various LLMs, including Baidu’s ERNIE Bot, to teach military AI to confront unpredictable human enemies.
Go deeper with GlobalData
Baidu clarified that ERNIE Bot is publicly available and accessible to all users, refuting claims of any specific collaboration with the authors or institutions involved in the research.
The controversy originated from an article by the South China Morning Post, which initially suggested a physical link between the military AI system and Baidu’s ERNIE Bot. However, Baidu’s statement highlighted that the media outlet later corrected its report, clarifying that there was no direct connection.
Baidu’s official statement on the matter reads, “We have no affiliation or other partnership with the academic institution in question. We have no knowledge of the research project, and if our large language model was used, it would have been the version publicly available online.”
The military AI, developed by the PLA’s Strategic Support Force, reportedly utilised ERNIE Bot and iFlyTek’s Spark, similar to OpenAI’s ChatGPT, in experiments to enhance its ability to understand and respond to human adversaries.
Baidu, however, emphasised its lack of knowledge about the specific research project and affirmed that, if ERNIE Bot was used, it would have been the publicly available version.
The PLA’s initiative to integrate commercial language models into military AI has raised concerns about the potential implications. The military AI reportedly employed LLMs to convert sensor data into descriptive language or images, with the entire process automated and devoid of human involvement.
While the researchers believe the project could benefit both humans and machines by enhancing decision-making and combat knowledge, critics warn of potential risks.
In response to the controversy, Baidu’s stock experienced a significant decline, with shares slumping 12% in Hong Kong. The company issued a statement addressing the situation, emphasising its commitment to operating within legal and regulatory frameworks.
