huawei's tao jingwen: ai transformation needs to organize data around scenario processes and it should be implemented layer by layer
2024-09-22
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
cover news reporter meng mei and fu wenchao
on september 19, at the huawei connect conference 2024, huawei director and president of quality process it tao jingwen delivered a keynote speech entitled "digital transformation and intelligent reform to achieve continuous high-quality development of the enterprise", sharing huawei's digital transformation process and its own practice of comprehensive intelligent upgrades.
tao jingwen said that if enterprises want to make good use of ai, they must first upgrade their ai data governance system. the large amount of structured data accumulated in the information age and the large amount of documents generated by past enterprise operations cannot be directly recognized and learned by ai. "in order to deeply integrate the ai model with the data tool chain, we must add a layer of ai security governance and ai digital production line on the basis of traditional data governance and data platform, and then build new ai services for enterprises based on huawei cloud's new ai computing power platform."
ai is a revolution. if enterprises want to turn ai into a service that can truly bring value to the business, they must deeply integrate it with the enterprise's processes, organization, it, data, and business scenarios. huawei has summarized a set of "three layers, five levels, and eight steps" methodology based on its own practice. the three layers are divided into redefining intelligent business, ai development and delivery, and continuous operation of intelligent applications. the "five levels and eight steps" are to start from the business scenario and guide the business along the business process, organization, company data, and ai applications on how to implement the enterprise's ai step by step. tao jingwen emphasized that not all scenarios are suitable for ai. for this reason, huawei has defined the "twelve questions" of ai scenarios as a key means of selecting intelligent business scenarios.
tao jingwen said that in the process of implementing intelligence, it is necessary to adhere to scenario-driven development to solve the company's massive, repetitive, complex and high-energy consumption problems. "multi-model-based multi-modal system engineering is the focus of enterprise implementation of ai."
for example, in contract scenarios, huawei digitizes objects, processes, and rules to achieve high-quality parallel processing of massive contracts, intelligent extraction and comparison of multi-language contract elements, and reduces risk operations from 2 hours to 5 minutes.
according to reports, in order to better integrate ai with enterprise business processes and data, huawei has built an enterprise ai digital production line. tao jingwen said that enterprise applications require more certainty in model results. in the process of model training, enterprise data sets are unique and professional data, information and knowledge, which are more professional than professionals. continuously improving data quality to a "textbook" level is the most effective way to deal with model uncertainty.
finally, tao jingwen concluded that transparency generates trust, and trust promotes sharing. huawei combines its own digital and intelligent practices with industry practices to develop a set of enterprise digital space governance models, unifying data and language, and drawing an efficient, intelligent, transparent and trustworthy digital governance blueprint for enterprises, helping enterprises achieve digital transformation and intelligent upgrades, and achieving sustainable high-quality development of enterprises.
