LLMs are extending the number of operations we can perform without conscious thought
LLMs as the cognitive-era cooking/washing machine
The future of technology with Large Language Models (LLMs) can be seen as an embodiment of A.N. Whitehead’s idea that “Civilization advances by extending the number of important operations which we can perform without thinking about them.” LLMs, like ChatGPT, are extending the number of operations we can perform without conscious thought.
The evolution of civilization has always been about offloading brain cycles to technology. Cooking is a prime example. Early humans spent significant time and energy hunting and gathering. The invention of cooking allowed us to pre-digest food, saving energy and freeing up time for other pursuits. This offloading of ‘digestive’ brain cycles to technology (fire) was a significant leap in our evolution.
Fast forward to the industrial revolution, we see another example with the washing machine. Before its invention, washing clothes was a labor-intensive task, consuming a significant portion of our time. The washing machine automated this process, freeing up more brain cycles for other tasks.
In both cases, technology took over tasks that were previously manual and time-consuming, allowing us to perform them without conscious thought. This is the essence of civilization’s advancement as per Whitehead’s idea.
Now, we’re seeing a similar trend with LLMs. They’re taking over tasks that require cognitive effort, like writing emails or coding, freeing up our brain cycles for higher-level thinking. This is the next step in our civilization’s evolution, offloading cognitive brain cycles to AI, just as we did with physical tasks to machines in the past.
In essence, LLMs are the cooking fires and washing machines of the cognitive era.
In this context, we can draw parallels between LLMs and theories from psychology: Kahneman’s System 1 and System 2. LLMs handle tasks quickly and without much conscious thought, akin to System 1. However, like System 2, there’s a need for human oversight and decision-making in the operations performed by LLMs.
A common mistake made by many AI startups is the attempt to completely replace human involvement, leading to a loss in quality. This approach overlooks the potential of an iterative process where human feedback is incorporated into the AI system in an optimization loop. This iterative process, akin to a feedback loop, allows for continuous improvement and refinement of the AI system, ultimately leading to a global maxima of human desire.
The optimal use of AI, and particularly LLMs, is not to replace humans but to augment human capabilities and reduce cognitive load.
The future of technology with LLMs seems to be moving towards a reality where humans live in a bubble of advanced AI systems, executing our goals and reducing our cognitive load. This aligns with Whitehead’s idea of civilization’s advancement, as we extend the number of important operations that we can perform without conscious thought.
Interestingly, this vision of the future also aligns with the principles of the Elon Musk Algorithm, a 5-step process used by Elon Musk to improve and optimize various aspects of his businesses. The final step of this algorithm involves automating tasks wherever possible, which is precisely what LLMs are designed to do. However, as we move towards this future, it’s crucial to remember the importance of human co-iteration and decision-making in the operations performed by these models.