数据工厂“去OEM化”的战略路径研究——以数据标注行业为例

Research on the strategic path of "de-OEMS" in data factories -- A case study of data labeling industry

摘要:

近年来，人工智能作为一个战略新兴产业已经成为我国“新基建”建设中的重要一环。人工智能的发展得益于算法、数据和算力三方面共同的进步。由此，对数据的需求正呈现爆发式的增长。大部分科技公司将其需要的数据转交给数据工厂，通过数据工厂的处理人员进行加工以节约成本，既推动了人工智能的发展，又促进了就业。数据标注行业也因其劳动密集型、代加工的特征而被印上了新时代“富士康”的标签。然而，由于该行业处于快速发展阶段，对技术更新速度要求较高、行业内缺乏统一的规范和标准，加之我国面临着人工成本日益上涨等问题，数据标注行业正面临着潜在的风险与危机。本文对我国数据标注行业进行探索与研究、与国外先进公司进行比较，并识别出该行业的潜在问题，通过构建转型升级路径实现数据工厂的“去OEM化”以提高数据处理的效率及质量，最终实现由劳动密集型产业向技术密集型产业的转变。

Abstract:

In recent years, as a strategic emerging industry, artificial intelligence has become an important part of China's "new infrastructure" construction. The development of artificial intelligence benefits from the common progress of algorithms, data and computing power. As a result, the demand for data is exploding. Most technology companies transfer the data they need to data factories, which are processed by a large number of data processing personnel to save costs, which not only promotes the development of artificial intelligence, but also promotes employment. Data labeling industry was also covered by the new era of “Foxconn” label because of its labor-intensive and OEM characteristics. However, due to the rapid development of the industry, high requirements on the speed of technological update, the lack of unified norms and standards in the industry, and the rising labor costs in China, the data annotation industry is facing potential risks and crises. By exploring and studying the data annotation industry and comparing it with foreign advanced companies, this paper constructs its transformation and upgrading path and realizes "de-OEM" to improve the efficiency and quality of data processing, and finally realizes the transformation from labor-intensive industry to capital-intensive industry.