Professional Shyogi (japanese chess) player asked about how he views use of AI in shyogi practices, and the AI software as a player.
In the interview, he mentions that “AI makes fewer mistakes, but when they do, AI has a greater degree of mistakes. Humans make mistakes, but I don’t think they make extremely simple mistakes. Basically I don’t know what AI is doing in the process. As with deep learning, the process is too large and difficult to understand. In other words, I think that the intelligence that humans have is to re-evaluate and re-define that part (process) in their own way, and identify fluctuations. For example, when AI makes a decision, human can evaluate its decision, or if it is questionable include the human’s thoughts and view-point in the process. This could improve the accuracy of the use of AI”
“I question our attitude to let us get closer to AI, instead of AI getting closer to us. In shyogi playing, as it is a technical problem, it is easier to play like AI than other way around, and I guess that is the reason for the recent trend of playing like AI. I personally do not like how it is, though.”
https://digital.asahi.com/articles/ASM5G7F3RM5GULFA04F.html?iref=pc_rellink