Easily Making Use of Such Technology Might Cause Extensive Harm

The use of conversational artificial intelligence (AI) capable of producing highly accurate sentences has been spreading. Easily relying on such tools, however, might undermine people’s ability to think. The ways to make use of the technology should be explored carefully.

ChatGPT, a conversational AI model developed by a U.S. startup, is designed to respond to questions typed online in natural sentences as if written by a human. Since ChatGPT was released free for public use last November, the number of users has reportedly exceeded 100 million worldwide.

There are hopes that it can be used in business. For example, when asked, “Can you think of a slogan to promote a hot spring resort area?” ChatGPT can instantly offer an idea. It can also create various business-related documents.

As such AI tools can also interact with clients, Bengo4.com, Inc., which operates a legal consultation website, said it plans to launch consultation services using AI this spring.

While AI has the potential to improve work efficiency and create new business opportunities for companies, there are many concerns about its wider use.

The content of sentences created by AI is not always accurate. Because the sources from which the AI has learned are unspecified, there is also the risk that the text and other content are copyrighted works of other parties that are then being unknowingly used for products and services. Another risk is that data input to the AI might be leaked externally.

The educational field faces an even bigger problem. For example, given the instruction “Write a book review on Osamu Dazai’s ‘No Longer Human’ within 800 characters,” a response can be immediately generated, so children might take advantage of this. University students might also misuse the technology for their academic papers.

If such use becomes widespread, people could lose their ability to compose sentences using their own thoughts. Blind use of such AI tools is extremely dangerous and must be curbed to a certain extent.

In the United States, ChatGPT has indeed shown its ability to produce academic papers at the level of those created by university students, prompting New York City to block access to ChatGPT from devices and networks in schools under its jurisdiction.

There have also been moves in the United States calling for people to sign online petitions to demand that research institutions stop the development of AI, claiming that it poses serious risks to humanity.

While examining possible risks from various perspectives, Japan also must urgently consider laws and regulations.

In the Diet, a House of Representatives member from the Constitutional Democratic Party of Japan asked Prime Minister Fumio Kishida a question created by ChatGPT about measures against infectious diseases. The lawmaker also prepared an answer to the question using ChatGPT and compared it with the prime minister’s answer.

It is undesirable to use and praise AI in such an easy manner. It is important to discuss AI-related issues in the Diet, such as how to regulate the technology.

(From The Yomiuri Shimbun, April 2, 2023)