在进行数据治理的过程中，遇到了两个挑战：第一，数据多而分散，质量参差不齐，数据治理难度大；第二，不同粒度的数据难以有效融合，数据价值被低估。而这些挑战一直会存在，不断出现，主要三个方面的现实原因。
认知问题：目前对数据治理的认知存在误区，很多人将其等同于传统的数据清洗，认为是技术性问题，缺乏对其重要性的认识。
当前，数据治理还是非常传统，对数据治理不重视，而不重视的原因是企业对数据治理的认知是有问题的，还是把数据治理等同于传统的数据清洗。另外，数据治理应该包含安全合规、清晰透明、公平多样、高质高效等多个方面，而不仅仅是技术层面的处理。因为，数据治理单纯依靠技术无法解决，其中管理和沟通占据70%以上的工作，特别是传统行业大型集团型企业，IT系统差异大，业务和组织复杂度高，因此需要数据治理人员加倍耐心与各层级各条线沟通，深入理解业务，以布道者和服务者的姿态逐步推进数据治理的深入。
原生驱动力不足：数据治理往往因为外部压力（如成本控制）而启动，缺乏内在的原生驱动力，使得数据治理难以得到有效推进。
目前有的企业数据治理做的不好还是因为“不够痛”，第一是数据量太少，根据经验，一般企业只要数据用起来，数据的增长速度在一年涨一倍、两年涨三倍、三年涨七到九倍的范围。一旦数据开始增长，会出现存储费用高的问题，这时企业会意识到必须要做治理、降成本、做运维。
在公司里除非被迫，如管理层要求服务器成本必须降，才会有人去做这件事。这就变成了一种硬性指标，缺乏原生的驱动力，而是被动去干“脏话累活”。如果把数据治理等价于脏活累活，那这个数据治理怎么可能做的好呢？
另外，在实际项目中还存在这么个问题，比如老板给一个项目要求今天做完，那么员工一定不会想先去做数据，首先想到的是先做出一个能交差的结果。所以对数据的不重视是数据治理的一个最大问题。
企业历史数据的处理：历史数据的错误和问题难以丢弃，需要在治理过程中加以处理和纠正。
数据治理做不好，很多时候也是有心无力，尤其是大企业，其历史发展时间太长，以往的一些代码产生的问题数据没办法丢掉，只能在其错误数据基础上不停迭代，但这种情况也应该是有办法做好数据治理的。
综上所述，数据治理面临的挑战和机遇涵盖了认知问题、原生驱动力不足、数据治理的范围和目标、企业历史数据的处理、数据封闭性问题、数据治理的层次结构、法律和合规问题、大模型与数据治理的结合、数据价值与制度设计、人才培养和研究合作等多个方面。每个方面都需要综合考虑，才能有效推进数据治理的发展。
翻译：
Talk about the real-world challenges of data governance
In the process of data governance, there are two challenges: first, the data is numerous and scattered, the quality is uneven, and the data governance is difficult; Second, it is difficult to integrate data of different granularity effectively, and the value of data is underestimated. And these challenges will always exist, and continue to appear, mainly for three practical reasons.
Cognitive issues: The current perception of data governance is flawed, and many people equate it with traditional data cleaning, thinking that it is a technical issue, and lack of understanding of its importance.
At present, data governance is still very traditional, and the reason for not paying attention to data governance is that the enterprise’s cognition of data governance is problematic, or the data governance is equivalent to traditional data cleaning. In addition, data governance should include security compliance, clarity and transparency, fairness and diversity, high quality and efficiency, and not just technical processing. Because data governance cannot be solved solely by technology, among which management and communication account for more than 70% of the work, especially in large group enterprises in traditional industries, with large differences in IT systems and high business and organizational complexity, data governance personnel need to be more patient to communicate with all levels and lines to deeply understand the business. Gradually promote the depth of data governance as a preacher and a service provider.
Lack of native drivers: Data governance is often initiated due to external pressures (such as cost control), and the lack of intrinsic native drivers makes it difficult to effectively promote data governance.
At present, some enterprise data governance is not good because of “not enough pain”, the first is too little data, according to experience, the general enterprise as long as the data is used, the growth rate of data in a year doubled, two years increased by three times, three years increased by seven to nine times the range. Once the data starts to grow, there will be high storage costs, and then enterprises will realize that they must do governance, reduce costs, and do operations.
In a company, no one will do this unless forced, such as management, to reduce server costs. This has become a hard indicator, lack of native driving force, but passive to do “dirty work.” If data governance is equivalent to dirty work, how can this data governance be done well?
In addition, there is still such a problem in the actual project, for example, the boss gives a project to be completed today, then the employee must not want to do the data first, the first thought is to make a result that can be handed over. So the lack of attention to data is one of the biggest problems in data governance.
Handling of historical data: Errors and issues with historical data are difficult to discard and need to be addressed and corrected in the governance process.
Data governance is not good, many times also have the heart to be powerless, especially large enterprises, its historical development time is too long, some of the past code generated problem data can not be lost, can only iterate on the basis of its wrong data, but this situation should also be a way to do a good job of data governance.
In summary, the challenges and opportunities for data governance cover cognitive issues, lack of native drivers, the scope and objectives of data governance, the processing of historical enterprise data, data closing issues, data governance hierarchy, legal and compliance issues, the combination of large models and data governance, data value and institutional design, talent training and research collaboration. Each aspect needs to be considered in order to effectively advance the development of data governance.
