China begins testing AI chatbot for brain surgeons in hospitals

State organs are joining private Chinese firms in developing homegrown AI in the mold of OpenAI’s ChatGPT.

China is testing an AI assistant for neurosurgeons at seven hospitals in Beijing and other cities in coming months, one of many initiatives the government is backing to try and harness the technology.

A Hong Kong-based agency under the Chinese Academy of Sciences — the country’s foremost state-backed scientific institute — on Monday introduced an AI model based on Meta Platforms Inc.’s Llama 2.0. Researchers trained and fine-tuned the model with papers, medical journals and manuals to act as a surgery consultant of sorts for doctors, said Liu Hongbin, the center’s executive director.

State organs are joining private Chinese firms in developing homegrown AI in the mold of OpenAI’s ChatGPT. The Chinese Academy of Sciences’ TaiChu model was among the first batch of services that won approval for public rollout in August. The technology is regarded as harboring the potential to revolutionize fields from diagnostics to personal consultations.

The Hong Kong-based Center for Artificial Intelligence and Robotics employed about 100 graphics processor units to train its healthcare-focused model, split evenly between Nvidia Corp.’s A100 high-end chips and Huawei Technologies Co.’s Ascend 910B, researchers told reporters Monday.

They hope the AI bot — dubbed the CARES Copilot 1.0 — will answer questions with citations based on more than a million academic records. It should also be able to process diagnostic data such as MRI, ultrasound or CT scans, as well as images, text and audio, they said. Eventually, the researchers envision the AI taking a more active role — warning doctors from pursuing risky procedures, for instance.

“There are obstacles including restricted computing power due to the banned access to Nvidia’s advanced chips,” said Feng Ming, chief physician at the Peking Union Medical College Hospital’s neurosurgery department, which took part in the model’s development. “However, we can develop a vertical model with our own characteristics with more high-quality data from top hospitals in the mainland, which is not available for OpenAI and many domestic private companies.”