Finding the optimal experience of AI: The journey of ZJU student CHENG Xinyu

At first, artificial intelligence (AI) seemed like a mystery to CHENG Xinyu. However, through lines of code and countless tests, he gradually cleared his path. Today, Cheng is a student in the class of 2022 in the Mixed Class of Chu Kochen Honors College at Zhejiang University (ZJU), majoring in Computer Science and Technology, dedicated to finding the "optimal experience" of AI.

Cheng's affinity for computer science began in high school, where he was a frequent participant in informatics competitions. After entering ZJU, he chose computer science without hesitation. His interest in AI was sparked during a term paper presentation for the "Mixed Learning Seminar" course. Later, he joined the Student Research Training Program (SRTP) focusing on large language models. Although he initially struggled, spending days in the library reading literature and debugging code, he gradually found his footing. In the summer of 2024, Cheng participated in an exchange program at the University of Oxford, earning an A+ grade and a scholarship, which further broadened his academic horizons.

Cheng's academic journey is also marked by competitive programming. He and his teammates spent nights preparing for the International Collegiate Programming Contest (ICPC), ultimately winning a silver medal in the Asia Regional. "Striving for the goal and the process of hard work are what matter most," Cheng recalled. The experience honed his ability to break down complex problems and collaborate under pressure, preparing him for future research challenges.

His research breakthrough occurred at the State Key Laboratory of Computer Aided Design and Computer Graphics. Under the guidance of Prof. WU Yingcai and Dr. WENG Di, Cheng co-authored a paper as a co-first author, which was accepted by the ACM Symposium on User Interface Software and Technology (UIST), a top-tier conference in human-computer interaction. Their project developed a dialogue model designed to generate data processing scripts, helping users modify code through human-computer interaction. This experience shifted Cheng's perspective from asking "what AI is" to "exploring what AI can do."

For Cheng, the most fascinating aspect of AI is its ability to process human emotions and ambiguity. Looking ahead, he aspires to become a computer science researcher. "I want to design practical, pioneering solutions that address real-world problems," he said. Just like an AI model iterating its algorithms, Cheng continues to take small steps toward a larger future.

Adapted and translated from the article written by Chen Siqing
Translator: JIANG Chenqi ('27, Structural Engineering)
Photo: the interviewees
Editor: JIANG Chenqi, DING Chenwei