Kenji Morishita (Executive Officer and VPoE) of PKSHA Technology, a company that develops AI services, has pointed out that there are cases in corporate generative AI implementation where the effects obtained do not match the costs invested.
NewsPKSHA TechnologyKenji Morishita
Determining the Cost-Effectiveness of Generative AI Implementation: Insights from PKSHA Technology's Kenji Morishita
This article is a translation. Read the Japanese original
Morishita stated that when proceeding with social implementation, it is crucial to determine whether generative AI is truly necessary from a cost perspective. For example, he noted that companies should judge whether investing vast sums to implement AI is appropriate compared to the cost of hiring humans skilled in natural language processing.
He also mentioned the strategic selection of technology based on the task. In cases requiring high-speed processing and the handling of massive amounts of data, such as credit card fraud detection, he argued that machine learning is more advantageous than generative AI. He cited examples where mathematical optimization was utilized instead of generative AI for tasks like shift scheduling, where results must be free of fluctuations and based on strict rules.
On the other hand, he highlighted the strengths of generative AI, such as the ability to operate using natural language without specialized knowledge and the capacity to read data that is difficult for machine learning to handle—such as handwritten documents or the layout of diagrams—in a "human-like" manner. Morishita explained that the key to implementing AI systems is the extent to which they can be adapted to the specific characteristics of the business operations.
Sources
- “割に合わない”AI導入、どう回避？ キホンを歴14年のPKSHAに聞いた （ITmedia AI+, 2026-09-10）