Software Engineering in the Era of Intelligence: the correct posture of embracing large models

The large model technology represented by ChatGPT has brought tremendous impact to many fields including software engineering, and also caused widespread anxiety. In order to see a little direction in the fog, we have been discussing and thinking about "software engineering in the era of large models" based on various technical literature, practical sharing, and our own preliminary exploration. Embracing large models should be a correct and even necessary direction for both the academic and industrial communities of software engineering, but how to achieve systematic and comprehensive intelligent development of software still requires calm thinking, as well as many basic work that needs to be done. This report will share our preliminary understanding and prospects for future development directions.


From code-specific AI to code-integrated general AI

ChatGPT and GPT-4 have opened the door to general artificial intelligence. Code plays a crucial role in these basic models. This report will start with AI models designed for code tasks, then move on to foundation models that integrate code data, and finally introduce our TaskMatrix architecture proposed for building AGI. We have done a good job of open-sourcing all the work covered in this report, hoping to promote the development of AI for Code and explore key technologies for AGI with Code.


New Paradigm of Software Development under Software Engineering 3.0 下载PPT

1. From Software Engineering 1.0 to Software Engineering 3.0 (SE3.0) 2. The new form of SE3.0 3. The new development paradigm of SE3.0 4. How will programming unfold under the new paradigm? 5. How can enterprises better utilize the new paradigm? 6. Future prospects and challenges


Solution and Application of Huawei's Large Model

1. Industry Insights of AIGC for SE 2. Huawei's Large-scale Code Model Solution and Application 3. Key Issues and Technical Challenges of Large-scale Code Models 4. Opportunities and Prospects of AIGC for SE


Intelligent Code: Exploring the Path from Task-Specific Models to General Large Models. 下载PPT

In this sharing session, we will explore the exploration process and future trends of code intelligence-related tasks in the field of software engineering, including three main parts: (1) The exploration path on task-specific models: In this part, we will review several works on task-specific models, including their core ideas and achievements in solving specific programming tasks. (2) The exploration path of general models in the software engineering field: This section will discuss the requirements and preliminary explorations from dedicated models to general models. (3) The road to future exploration: In the final part, we will explore the possibility of combining both and look forward to future development trends in code intelligence research.


How does AI programming improve productivity

How does AI programming improve productivity
Round table host :Qianxiang Wang | Chief Expert of Huawei Cloud Intelligent Software Development
©开源中国(OSChina.NET) 深圳市奥思网络科技有限公司版权所有 粤ICP备12009483号