ELAU C400/10/1/1/1/00
After experiencing the frenzy of the big model trend at the beginning of the year, more and more practitioners realize that the commercialization of generative AI is not as smooth as expected.
In industries such as industry, finance, healthcare, and education, generative AI has enough room for imagination, but many companies remain wait-and-see when facing new technologies. Especially in traditional industries, some business owners are more willing to choose manual labor when facing the multiple-choice questions of generative AI and professionals.
The technical threshold is one of the main factors for enterprise users to be cautious about generative AI. These vertical industry enterprises have a deep understanding of their own industry, but developing AI applications from scratch is not their expertise. There are also data security and privacy issues in this process, and the challenges and risks behind it make it difficult for them to let go.
Cost is another factor that keeps enterprise users cautious. Most enterprises plan their annual IT expenditures based on their existing business, and the inherent IT infrastructure of the enterprise is difficult to support generative AI applications that require large computing power and storage in the era of big models, making it difficult to cope with the sudden increase in demand for generative AI business. In addition, with the impact of the global shortage of chips, the computational cost of training and reasoning continues to rise, making it difficult for enterprises to control the cost of using generative AI, and they can only wait and see.
Under the influence of multiple unfavorable factors, the commercialization of generative AI in vertical industries has been relatively slow, and some practitioners have even begun to question whether current generative AI technology has the ability to make money.
Actually, it’s not necessary. Difficulties and challenges are temporary, and generative AI will have a period of rapid implementation opportunities as long as it breaks through the bottlenecks of technology and cost. At present, the key to clearing the obstacles to the implementation of generative AI lies in reducing the technical barriers and costs of developing AI applications, and cloud computing enterprises represented by Amazon Cloud Technology have taken the lead in providing solutions.
Enable enterprises to acquire AI “superpowers”
In the process of implementing generative AI, a noteworthy phenomenon is that “cloud based” enterprises with relatively fixed IT infrastructure are more commonly trapped in technical and cost difficulties, and these enterprises are mainly traditional enterprises. In contrast, companies with higher levels of cloud usage have lower trial and error costs.
The main reason behind the differences lies in the accumulation of technology. In the process of enterprise digitization, a higher degree of cloud usage also means a higher acceptance of new technologies and a greater accumulation of various technological capabilities, while traditional enterprises, mainly “under the cloud” enterprises, are the opposite. Developing generative AI applications requires more complex technical capabilities than going to the cloud, and traditional enterprises with weak technological accumulation face more challenging challenges.