As general large models flood in, artificial intelligence is accelerating into the deep waters of hard technology. On January 25, Shanghai Jiao Tong University officially launched a fully self-developed vertical large language model for the optical field - Optics GPT (Optical Large Model). This achievement not only marks an important breakthrough in the integration of "AI + hard technology" in China, but also equips the Chinese optical industry chain with a smart "digital brain" through precise professional insights.
If ChatGPT is an all-knowing "scholar", then Optics GPT is a "senior specialist" who has spent years in the laboratory and is full of knowledge. The "Light Future" project team at Shanghai Jiao Tong University did not take the shortcut of "old bottle with new wine", but instead let the model grow naturally from massive professional data on optical communication and optical design. It can deeply understand complex optical principles, acting as a 24/7 available "virtual optical expert", providing highly reliable intelligent support for research, design, and teaching.
This fully self-developed domestic large model has four core features: "light deployment, high cognition, strong application, and full controllability". Its emergence is making complex optical education accessible: dull physical formulas are transformed into intuitive visual demonstrations, and abstract theories become vivid through interactive Q&A. For researchers, it is also a versatile assistant, capable of quickly sorting through vast amounts of literature, and even inspiring innovative solutions in complex simulation calculations.
More dramatic changes have occurred at the forefront of the global computing power competition - optical module calibration. With the explosive demand for GPU interconnection, each outgoing optical module requires a significant amount of manual debugging of underlying code, making the testing phase a bottleneck for production capacity. Yi Li Lin, a specially appointed professor at the School of Integrated Circuits at Shanghai Jiao Tong University, pointed out that the involvement of Optics GPT can realize automated testing and calibration, greatly shortening the time to market. From fault prediction in computing centers, to intelligent interaction with high-end instruments, to autonomous optimization in laser manufacturing, the reach of the large model has extended to every capillary of the industrial chain.
With the simultaneous establishment of the "Optical Large Model Academic Ecosystem Alliance" and the "Industrial Ecosystem Alliance", Optics GPT is trying to break down technical barriers between optics, electricity, and algorithms. As Tang Xiongyan, vice president of the Research Institute of China Unicom, stated, this is not just an upgrade of single-point technology, but a revolution in the growth model of the entire optical communication industry. In the future of light and shadow, this domestic large model is reshaping the intelligent foundation of China's optical industry with its precise professional depth.
