Currently, AI cannot completely replicate a hand-painted original artwork in a short period of time. Let me give you an example. The other day, I conducted an experiment. I had a painting that depicted a dialogue between two versions of myself: the real-world me and the me in a dream, which symbolizes “death” as the soul. I trained an AI model for about three days, attempting to capture the essence of this painting. I inputted keywords, and the resulting artwork depicted a conversation between myself and a skull.
This is where machines cannot replace humans. AI, based on information retrieval, matched keywords such as “death” and “soul” with an automatic association to a “skull,” but it struggles to comprehend the Eastern aesthetic. Of course, if we continue to invest time in training the AI model, not just three days but three months or even longer, it is possible for AI to achieve better results. However, the time investment required may not be worthwhile.
The subjectivity and agency in artistic creation are strong. We can instruct AI to create paintings in different genres and styles, but this still relies on “human ideas.” Whether imitating past artistic styles or the current mindset of the creator, they all originate from human thoughts. At present, machines find it difficult to generate original impulses. For example, Raphael’s painting “The School of Athens” depicts 57 renowned scholars, representatives of different philosophical schools, engaging in equal and elegant discussions within classical architecture. They walk, talk, argue, calculate, and contemplate, each with a unique posture. If we were to input enough keywords, including the school of each person, their life stories, and detailed movements, I believe AI could replicate it. However, the ideas it generates would still be Raphael’s ideas.
Why did Raphael paint this gathering of 57 scholars? It was his original idea. Some argue that given enough time and extensive training, machines could also generate original impulses. However, that is not the focus of our current discussion.
Current AI technology excels in replication but falls short in terms of creativity. It may become more creative in the future, but for now, it cannot replace human creativity and imagination.
Therefore, it is foreseeable that in the AI era, jobs with low technological content, low creativity, and high replicability may be replaced. Jobs that involve creativity, imagination, emotions, and spiritual talents, such as art, are less likely to be replaced.