[2026 Davos Dialogue] Yuval Harari's Warning: When AI Takes Over Language, Human Identity Will Collapse

Whether discussing LLM, AGI, ASI, superintelligence, or any other terms, the first question that must be defined is 'What does Thinking mean?' Since René Descartes' declaration 'I think, therefore I am,' modern philosophy has believed that Thinking is a capacity that defines humanity. However, now, hundreds of years later, we are facing challenges to our understanding of Thinking from LLM (AI). Until now, there was a clear distinction that Thinking was a unique human ability, as nothing comparable existed. Since we have never encountered life forms elsewhere, to our knowledge, only humans have the capacity for Thinking. Therefore, until now, the somewhat unclear definition of Thinking was not a significant issue. However, with the emergence of LLM, we have entered a stage where we need to redefine the meaning of Thinking. LLM is a system that generates language by probabilistically predicting the next token rather than 'understanding' sentences. If Thinking is merely arranging words and language tokens in order, then LLM can be seen as thinking much better than most humans. As this language, Text, begins to be dominated by LLM, it will extend into law, literature, religion, and more, starting from simple website chatbots and beginning to influence human culture and spirit.