The evolution of artificial intelligence has gradually entered a new phase after the AI representatives led by ChatGPT shocked the world.

Since ChatGPT went viral, Microsoft has introduced open AI into its search engine, Google has released its own conversational bot, and Baidu, 360 and others are coming out with their own products.

What are the benefits and threats of AI, which is beginning to enter the linguistic and creative realms that are fundamental to human communication? The Nihon Keizai Shimbun interviewed Hiroaki Kitano, managing director and chief technology officer of SONY.

I was already aware of the technologies and mechanics that formed the basis of ChatGPT. In this sense, it is not surprising.

The question for the future is how to use it. Internet search (by Microsoft and others) will undoubtedly offer new models by interacting with very large databases and Information chunks.

Historically speaking, the development of AI is entering an important phase. According to current research, the AI (which became the basis of ChatGPT etc.) is trained to the power of 10 to the power of 22 to 24, and its performance increases dramatically. Although the improvement is weak below a certain amount of computation, once it exceeds a certain amount, it will make a leap to a higher stage.

Much of what we do on a daily basis may gradually be replaced by AI. There are also ongoing efforts to get AI to use tools such as software. It is not a scene from 2001: A Space Odyssey, but if AI can use tools, it will be a turning point. In the next three to five years, various attempts may accelerate progress.

From SONY group point of view, sooner or later will be in business to use. AI for "basic models" (advanced AI for generating articles and images, etc.) will also be developed. However, it is important to note that we are a creators-oriented company and will provide creators-oriented technology.

We will not develop AI that tramples on the rights of creators because existing AI generating images are being sued for copyright issues. It is important to develop fair, transparent, and accountable AI.

It may not be clear what SONY is doing in AI, but the company is at least deep in gaming. In February last year, for example, SONY developed an artificial intelligence (AI) that can beat a human "master" in a racing game. SONY Chairman and President Kenichiro Yoshida said it marked "a great leap forward in AI."

In 2019, Sony also created a Sony AI division, led globally by Hirohiko Kitano, president and CEO of SONY Computer Science Laboratories, and led in the United States by Peter Stone, a computer science professor at the University of Texas at Austin.

According to IT Home, Kitano is a former director of Sony CSL LABS and has been working on SONY's core business since the mid-1990s. SONY's first artificial intelligence product, Aibo, was born in the lab. Kitano is also the founding chairman of the RoboCup Federation, which aims to accelerate advances in artificial intelligence and robotics by developing a team of fully autonomous humanoid robots and winning the World Cup by 2050.

Foreign media have previously reported that Sony AI will have three core areas in the future: first, games, using AI to assist developers in design, testing, analysis and other work; One is imaging and sensors, which will continue SONY's legacy in the hardware market for another second. Finally, it is about food. It has jointly developed a cooking robot equipped with artificial intelligence with Carnegie Mellon University to realize cooking, serving and delivering food in one package.