I am reading Markus Gabriel’s new book, “The Sense of Thinking/What Thinking Means,” originally titled “Der Sinn des Denkens” in German, which translates to “The Sense of Thinking” in English. This dual-meaning title refers both to the physiological and bodily act of thinking, as more than just logical information processing, and to what it means to think in the age of AI and robots. While still working through the book, I would like to summarize my current understanding based on his arguments.
It is unlikely that AI will ever completely replace human intellect and dominate the world. We often mistakenly view thinking as merely a process of information handling. From my own use of AI, I have learned that it excels within a limited set of information in a controlled environment. This helps us grasp the insurmountable barrier between human thinking and what appears to be AI “thinking.”
Consider a scenario where one thinks about what to eat for lunch tomorrow. Based on experience, we know the reasonable scope of considerations, such as recent meals or budget-friendly restaurant options that fit with other plans and geographical limitations. If one lives in Hakodate, there’s no need to consider the Osaka subway schedule, the weather in Saudi Arabia, or chai prices in India. We inherently exclude such irrelevant information without explicit directives, rapidly defining the logical scope for our thoughts.
Thus, the act of human thinking begins with preset initial conditions that frame the logical process. When AI provides a seemingly appropriate response to “What’s good for lunch tomorrow?” it’s because the training data already contains past data defining such queries, with necessary conditions pre-assumed.
In contrast, AI struggles with undefined problems lacking explicitly related information. If asked vaguely, AI might fabricate responses—a phenomenon known as hallucination. AI can’t process numerical data meaningfully without specific instructions like “Plot the trend of basic statistical measures over time on a line graph.”
If AI appears to understand and engage with vague instructions like “make it nice,” it’s merely translating the colloquial meaning of “nice” into a specific processing command based on massive data from online chats and discussions. But if “make it nice” is meant to include ukiyo-e illustrations in the output, AI will likely fall short.
The book also touches on the halting problem, illustrating that no logical system can eliminate self-referential paradoxes like the Cretan paradox. Essentially, no perfect system can be created because it’s impossible to know beforehand whether a program will complete without issues.
Even rudimentary thought experiments confirm this. System development often stalls because it’s impossible to account for every potential event in advance, such as misinputs due to case sensitivity or invisible spaces, which can lead to unintended behaviors. While regular expressions and other screening measures are implemented, they can’t catch every bug.
Most processes that appear digitally managed are actually running smoothly due to external adjustments within certain limited conditions. It’s nearly impossible to incorporate all externalities into a system and create a perfect system—something anyone slightly familiar with computers would know from experience.
Thus, the operation of any system relies on humans preselecting events under certain conditions. Digital information processing systems can only solve problems explicitly defined with sufficient training data. Defining a problem, an act unique to human thought, is because thinking is fundamentally a sensory process, as argued by Markus Gabriel.
While AI can process information, recognize patterns, and execute tasks, problem-generative thinking remains uniquely human. This realization prompts us to maximize the benefits and efficiencies of AI development while emphasizing the importance of nurturing human cognitive abilities. More than solving individual problems, discovering and defining what constitutes a problem—akin to ancient Greek philosophical thought—will become increasingly crucial.