Recently, I encountered a statement from a professional shogi player, suggesting that AI proposes moves humans wouldn't consider. Although I'm not deeply familiar with shogi, I understand that there exist traditional "bad moves" embedded in its culture. Pursuing such moves might lead to admonishments from mentors. Yet, AI, devoid of these biases, simply identifies the best move for a given situation.
Despite the rigid rules of shogi, AI's selections might appear revolutionary to seasoned players. These unique moves, while not novel in essence, might be viewed as 'discoveries' or even acts of 'creation' by humans.
This concept could also relate to Chat GPT's answers. While humans operate with inherent biases, AI, though potentially influenced by its training data, isn't as heavily swayed by them. Hence, AI's responses often feel refreshingly innovative to us.
