07 Nov 2025
As China rapidly advances in artificial intelligence to boost economic growth and transform industries, some scientists are working to unleash the technology’s potential to promote social good.
Su Jionglong, deputy dean of the School of AI and Advanced Computing at Xian Jiaotong-Liverpool University, researches AI applications aimed at reducing discrimination, providing emotional support and making faster medical diagnoses.
He said he was establishing a start-up with his students, who developed an AI-powered software that translated written text to and from sign language.
“Limitless Mind is an inclusive communication platform to help overcome barriers between people who use sign languages and those who do not,” Su said.
“The lightweight proprietary model could be used as a mobile app or be installed on smart glasses,” he said, adding that virtual avatars would perform sign language motions or text would be shown in real-time.
Su said the team had been in talks with local governments and industrial parks, which were keen to provide financial support for products to be launched.
“We are innovating these technologies to meet real demands in society,” he said. “Disabled students can learn better in class, patients can have more succinct and meaningful conversations with doctors and workplace discrimination can be minimised.”
The next concepts Su and his students are attempting to bring to life are platforms that translate lip movements into text and brain-computer interface technologies that convert brainwaves into text, enabling control of driverless vehicles.
By 2060, more than 240 million people in China are projected to have moderate to complete hearing loss, doubling from 2015, according to a study published in the peer-reviewed Chinese Medical Journal in January.
“As the population of China increases and ages over future decades, the need for hearing care will increase,” the researchers in Beijing said.
Su said his educational background in mathematics, tatistics and engineering had prepared him for aspects of AI such as mathematical modelling, data analytics and pattern recognition.He previously worked in Britain at Warwick University and University College London, as well as in Kazakhstan at Nazarbayev University, where he was head of the mathematics department, before returning to China in 2014.
Two years later, AI became his research focus, in addition to his theoretical studies, such as the mathematical principles behind financial technology.
Su said one advantage of developing AI technologies in China was the vast data sets available to academics, such as medical data from hospitals which are “generous” in providing information to academics.
His team is working with the Shenzhen-based Mind with Heart Robotics, which develops electronic pets and childlike humanoids as emotional companions and psychological monitoring systems.
These robots could help support autistic children who have difficulties expressing their emotions, Su said.
“Caretakers might find it challenging to distinguish their feelings. Machine learning can classify emotional states in real time, enabling caretakers to respond appropriately,” he added.
He said training medical imaging AI needed a substantial amount of data, citing examples from his projects, including one that pre-screens symptoms using facial photos to detect unusual puffiness, swelling or discolouration that could indicate certain diseases.
“We each have intra-variability, meaning people have different skin colours and varying degrees of facial smoothness.
This variability has to be removed for the AI to accurately classify symptoms,” Su said.
“For an AI to analyse multiple symptoms simultaneously, an exponentially large amount of labelled data is needed for training, with each photo annotated to tell the AI if certain symptoms are shown.
“Doctors sometimes come to us, initiate collaborations based on high-quality data. Top-tier hospitals store data in ahighly consistent and systemic way. Doctors use the same machines for data collection, reducing the variability within images and data because they were captured by different machines and operators.”
He said China’s strategic planning to become a world leader in AI by 2030 was favourable to researchers.
“National support is evident through funding that is strategically focused on areas like chips, given geopolitical tensions with the US surrounding semiconductors. There is also significant investment in generative AI and superintelligence,” Su said.
“Local governments, such as in Hangzhou and Shenzhen, have policies to establish AI hubs and attract talent and innovation.”
AI development in China has made significant breakthroughs in recent years, with start-ups such as DeepSeek gaining global attention by launching two advanced open-source AI models at a much lower cost than larger US tech firms.In April, President Xi Jinping said that despite some progress, much work still needed to be done in China to “achieve self-reliance” in AI. He said the country should “continue to strengthen basic research and concentrate on conquering core technologies such as high-end chips and basic software”.
This month, Tsinghua University in Beijing claimed the top spot in computer science across several major global rankings, signifying a shift in a field once dominated by American universities.
While China excelled at developing applications and turning research into products, Su said the US advantage in AI lay in its strong fundamental research, which surpassed that of China.
“To encourage greater creativity from institutes, some national funding schemes could be decentralised to allow for more flexibility at the provincial level, adding to the top-down approach focused on strategic goals,” he said.
“A new peer-review system for grant applications could also promote innovation, especially in fundamental technologies.”
Su said exchanges and collaborations between academia, industry and business were essential for driving innovation and addressing real-world challenges.
“A researcher focused solely on publishing in top academic journals may become disconnected from companies and irrelevant to the industry,” he said.
“Supporting robust basic research, such as mathematics, is just as important as encouraging risk-taking, which will help scientists commercialise their research more easily and learn the ropes of the industry.”
07 Nov 2025