At a recent AIGC-themed forum co-organized by Fudan University School of Management, Professor Lu Xianghua from the Department of Information Management and Business Intelligence delivered a keynote speech on “AIGC and the Era of the Intelligent Enterprise”, sharing her latest research and insights on the new possibilities that AIGC technology will bring along to corporate management, and what enterprises can do to harness this technology so as to achieve the full potential of digital transformation, etc.
1 Enterprise Intellectualization and AIGC Shares the Same Vision
Since the end of last year, development in the AIGC technology has been fuelling a new craze around the world. In terms of enterprise management, AIGC has had a great impact on information processing and distribution, human-computer interaction, business linkage, automatic optimization, etc.. So what kind of new possibilities will AIGC technology bring to the digital transformation of enterprises?
We usually divide the digital transformation of enterprises into three stages. In the early phase, enterprises mainly focus on ERP and CRM; next, they can move on to digitalization; then at last, intellectualization of the enterprise management is achieved with the data accumulated over time.
The essence of the enterprise intellectualization is to enable proactive prediction, proactive prevention, quick response, and automatic adjustment through the leverage of data, models, and computing capabilities. In this sense, intellectualization shares the same goal and infrastructure as the AIGC applications, which explains the enterprises' enthusiasm for this technology.
According to surveys conducted by the think tank EqualOcean, ten years ago, only 4.3% of enterprises were willing to implement digitalization within one year; whereas as many as 90.8% of enterprises nowadays are planning to develop AIGC applications within one year.
So far, the AIGC industry has not achieved a strong product-market fit. Now that the market of generative AI is entering “Act II”, as Sequoia Capital put it, we need to think more about how to implement the technology into specific applications of enterprises through large industry models and large scenario models.
2 AIGC Helps “Activating” RPA
AIGC applications have been developed in various fields, such as marketing, customer service, OA, finance, HR, R&D, supply chain, etc., although most of them are still in their early stages.
Meanwhile, the application of AIGC in smart customer service is relatively mature. Before the deployment of AIGC technology, more than 80% of requests had to be transferred to human agents; now thanks to AIGC chatbots’ strength in empathy, comprehension and logic, this number has been reduced to 40%.
With the improvement in internal knowledge base and information retrieval brought along by AIGC, new employees don’t have to rely much on instructors and orientations any more. For companies like Kidswant Children Products, which owns more than 200 marketing gadgets, AIGC has saved tremendous amount of time and energy for the employees and reduced the cost of training for the company.
Much more capable than conducting effective conversations, AIGC applications can also directly trigger other applications to complete certain tasks, fundamentally improving RPA (Robotic Process Automation) and making low-code/no-code programming possible.
3 AIGC Application Development Takes Patience
Development of AIGC doesn’t necessarily lead to layoffs. The technology replaces more tasks than positions.
AIGC doesn’t produce values automatically. The prompt engineering can be very essential to its effect. Those who claim articles, posters, plannings, etc. generated by AI are unacceptable often have very limited interactions with AI, or have not yet mastered the methodology for such dialogues. Naturally, AIGC application development requires a lot of patience and debugging, which underscores the necessity of large industry models and large scenario models.
Meanwhile, it’s important to note that not all enterprises need to develop their own large models. For most small and micro enterprises, adopting mature open source APIs or third-party softwares are often a more suitable solution.
