Against the tide: Two scientists who bet their careers on unfashionable questions

Xie Weidi was once so absorbed in video games that his academic performance in college suffered. Chen Peining fell short of admission to his dream university by just two points in the English section of his graduate-school entrance examination. Today both of them are among China's promising young scientists, pursuing research in fields that few others were willing to explore.

On October 31, at the 2026 World Laureates Forum (WLF) in Shanghai, the two were named among six recipients of the inaugural WLF Prize for Young Scientists, each receiving 1 million yuan ($149,430). The recognition marked not only what they had achieved, but also the promise of what lies ahead.

Chen is a professor of polymer science at Fudan University and develops flexible electronic materials and devices. Xie is an associate professor of artificial intelligence at Shanghai Jiao Tong University and co-founder and CEO of Guanyi Intelligent and works at the intersection of computer vision and the life sciences.

In interviews ahead of the forum, the two scientists returned repeatedly to a common theme: how they found their fields, and why they chose paths that few others were following.

A taste for the unexplored

Neither scientist followed a conventional path into research.

When Xie arrived at University College London to pursue a master's degree, he had modest ambitions. "I only wanted a master's and a job back home," he recalls.

But he soon found himself drawn to a field still taking shape. "Communications already had a map," he explains. "The path for 4G, 5G and 6G was clear. Computer vision was new, and full of unknowns."

He chose computer vision — and then took on a project that even his own field regarded with skepticism: training a model to recognize Arabic handwriting. Arabic handwriting presented distinctive recognition challenges because letterforms can change depending on their position in a word.

His adviser initially opposed the idea and was seldom available. Xie met with a postdoctoral researcher once every month or two, leaving him largely to work things out on his own. He taught himself linear algebra and machine learning through online lectures offered by MIT, completing the project independently.

When the results finally came in, his adviser described them as an unexpected gift and told him he had a natural feel for research.

That early experience eventually took Xie to Oxford's Visual Geometry Group, where he studied under Andrew Zisserman. Zisserman demanded that his students justify every word in a paper. Xie initially found the process exhausting. Over time, however, he became even more exacting than his mentor, debating individual words with him.

The discipline has stayed with him. He now runs his own lab with the same attention to detail, scrutinizing everything from the capitalization of a report title to the reasoning behind a model's design and the way its data is handled.

So has his instinct for the unexplored — what Xie describes as "taste": the ability to recognize which questions are worth devoting years to.

Chen's route into research was less direct. During his first two years at university, he had little idea what he wanted to pursue. He explored courses across different departments, including the humanities. It was not until his third year, when he entered a laboratory to help a senior student with experiments, that he discovered how much he enjoyed research. He subsequently applied to Fudan's polymer science program.

"My path as a student was not smooth, but I never gave up," Chen recalls.

He went on to enter Fudan's doctoral program, undertake postdoctoral research in Canada, and return to the university in 2017.

"Interest is the best teacher," he says. It helps researchers identify where to concentrate their efforts, he argues, and is the starting point for developing the kind of "taste" that guides scientific discovery. Universities, in his view, have a responsibility to give students the freedom to explore. "If I had not taken those courses or worked in a laboratory, I might never have learned what kind of research I prefer," Chen adds.

The hardest 10 percent

For Xie, a decisive shift came in 2022, when he chose an academic career, returned to China and redirected his research toward "AI for Science."

He set his sights on one of the field's most challenging applications: rare diseases.

A single rare disease may affect only a few thousand people. The resulting scarcity of standardized data makes it difficult to train AI models, while the small patient population offers little commercial incentive for pharmaceutical companies to invest in treatments.

"Choose the problems that rank in the top 10 percent for difficulty," Xie says. "That screens out 90 percent of the people. If you have the determination to step in, you are already in the top 10 percent."

The economics help explain why so few researchers and companies venture into this area. A pharmaceutical company that spends $500 million developing a treatment for just 5,000 patients is unlikely to recoup its investment.

"From a commercial angle, it is a losing business," Xie says. "From a human angle, how can you not do it?" he asks.

He says AI could change that equation. A single model, he argues, could potentially be applied to 50, 100 or even 1,000 diseases, dramatically reducing the cost of research for each one.

His team is developing a foundation model designed to capture the underlying processes of life: how genetic information flows from DNA to RNA to proteins, how these stages interact, and how they ultimately shape cellular function and human health. By modeling these interconnected processes, the researchers hope to better understand how diseases arise and progress, and then identify potential points of intervention.

The team began with rare diseases whose genetic mechanisms are relatively well understood. The team's longer-term ambition is to extend the approach to a broader range of conditions, including tumors, metabolic disorders and neurological diseases.

Xie does feel the pressure of competition, but it comes primarily from the gap between his team's capabilities and the world's most advanced AI models. Progress in AI depends heavily on concentrated talent and substantial resources, particularly data, computing power and funding. His team is still working to catch up with the leading models worldwide. The challenge is formidable, but so is his confidence.

For Chen, the challenge lies in developing flexible electronic materials and devices, technologies that could eventually take the form of a fabric capable of displaying images.

His laboratory has spent years trying to make that vision a reality. Producing the devices themselves is only part of the challenge as matching the resolution of conventional commercial displays has proven far more difficult than expected.

There were times when Chen thought the team had reached the limits of what it could achieve. Then a new finding would open up another possibility, gradually bringing the direction of the research into focus.

The lab now sees a viable route toward commercial-display resolution and has begun exploring the next frontier: flexible integrated circuits. The ambition is to give a fiber the information-processing capabilities of a chip.

"What kind of electronic device do we need in the AI era?" Chen asks.

He is pursuing the answer through interdisciplinary collaboration and hopes to work closely with industry partners to translate research into practical applications.

"If I could choose again, I would still choose this direction," Chen reflects.

What a university is for

The work of a scientist extends well beyond the laboratory. Both men now lead research groups of their own, and both have given considerable thought to what universities can offer that companies cannot.

For Chen, the answer begins with freedom.

Universities and companies share many characteristics, he says, but academic institutions have greater latitude to explore questions without immediate commercial pressure. More important, he says, is what students gain from their university education: intellectual breadth, curiosity, the ability to cross disciplinary boundaries and the confidence to tackle unfamiliar problems.

"I hope they become creators, not executors," Chen says.

A student who spends years exploring a subject, eventually commits to research and perseveres, he adds, is more valuable to him than a stack of published papers.

"Papers are the natural product of the research process," Chen notes. They document the work; they should not dictate its direction. What matters is whether the research addresses a genuine problem and advances the field. Research evaluation, he argues, should take a longer-term view, giving young scientists with genuine intellectual drive the freedom to venture into areas where few others — or perhaps no one — is working.

Xie shares that view, putting it more bluntly. To him, a paper is simply a written account of research that has already been completed, not a target around which researchers should design their work. The decisive factor is the choice of problem: identify the right question, pursue it rigorously, and the publications will follow.

For Xie, the value of an academic career is also rooted in the people universities bring together. Companies may be better positioned to scale existing technologies, but universities are where the next generation of innovators is trained.

Students arrive with considerable talent and ambition, and a teacher's responsibility is to help each discover a direction of their own. "Leading companies recruit the top 10 percent of talent from universities," he says. "Our job is to help the other 90 percent reach or even surpass that level."

This goal, Xie says, is best accomplished through example rather than instruction. Students learn what research really means by observing how their advisers work day after day, not simply by listening to lectures about scientific inquiry.

When interviewing prospective students, he asks them to explain their previous projects: how they formulated the research questions, how their ideas developed, and how they overcame the most difficult obstacles.

Research, he tells them, is built on sustained effort, repeated verification and years of exploration. He wants students to understand that it requires single-minded dedication and a willingness to accept mistakes — both their own and those made by others.

Students often ask whether a particular model will train successfully. "Maybe, maybe not," he tells them. "If not, we try another method." Exploration is never wasted.

He is also keenly aware of the pressure students face to follow the latest trends. In interviews each year for direct-entry PhD programs, he finds that 70 to 80 percent of applicants gravitate toward the same handful of fields: multimodal models, embodied intelligence and coding agents.

"In five years, when you finish, will that job still exist?" he asks them.

Rather than chasing the latest popular topics, he encourages students to look beyond the immediate excitement, discover what genuinely interests them and consider what they are prepared to pursue over the long term.

Xie puts the choice in more personal terms. "Find something you are willing to give your life to, and never regret it," he says. "Not finding that something, that is the real pity."