A Chinese quantum computing company has made a groundbreaking advancement in artificial intelligence, demonstrating that quantum hardware can significantly enhance AI efficiency. Origin Quantum, a startup based in Hefei, eastern China, has successfully fine-tuned a billion-parameter AI model using its proprietary quantum computer, Origin Wukong.
This achievement marks the world’s first real-world application of quantum computing in large AI model tasks. According to Chen Zhaoyun, a scientist from the Hefei Comprehensive National Science Center, this breakthrough shows that current quantum computers are capable of supporting the fine-tuning of AI models.
In their experiments, Origin Quantum managed to reduce the number of parameters in the AI model by 76%, resulting in an 8.4% improvement in training performance. This suggests that quantum computing can enable more efficient, lightweight AI models, offering a promising new approach to overcoming future constraints in computing power.
Fine-tuning involves adapting a general-purpose AI model for specialized applications by retraining it on specific data. For instance, a model like DeepSeek can be tailored for use in fields such as medical diagnosis or financial risk analysis.
The Origin Wukong, a 72-qubit programmable superconducting quantum computer, has been providing quantum computing cloud services since its launch. It has served over 23 million users across 139 countries, successfully executing 350,000 quantum computing tasks.
This milestone by Origin Quantum signals a significant step forward in the integration of quantum computing and artificial intelligence, potentially transforming how we approach complex computational challenges in the future.
