Abstract

Brainstorming, personnel evaluation, role-playing, demand forecasting, internal knowledge search...
What is the line between "jobs that can be entrusted to generative AI" and "jobs that should not be entrusted to generative AI"?

We're increasingly hearing requests like: "I want to automate email newsletter distribution." "I want proposals created based on internal documents." "I want a system that can respond to customer inquiries 24 hours a day."

AI has already become ingrained in our lives, streamlining everything from daily life to work. However, when actually implementing it, there are many situations where people are confused, saying, "It's not as accurate as I expected" or "It doesn't work as expected." Behind these situations, in most cases, is excessive expectation and misunderstanding about AI.

Where is the line between "jobs that can be entrusted to generative AI" and "jobs that should not be entrusted to generative AI"? Based on use cases such as brainstorming, document summarization, sales AI agents, and email newsletter generation, an up-and-coming data scientist explores the potential and limitations of generative AI and LLM, as well as points to note when using it in business.

Author’s Information

Representative Director of NIRO/Director of Akihabara Republic.

Graduated from the Faculty of Engineering at Kumamoto University. Data scientist. His mission is to implement cutting-edge technology in society. He translates specialized technology into "language that people can understand" and provides support to companies, mainly in the fields of AI and data utilization.