Thoughtless and Shameless
Much of my thinking has been outsourced to AI. Perhaps this is the ethos of the new era we have entered. There is joy in being able to know more and do more, yet… for some reason, the joy of achievement doesn’t last very long. Building things with AI, doing research, studying, writing papers—I have tried plenty of the things I used to want to do. I marvel at the water gushing down from a waterfall. Greedy to catch it all, I thrust a ladle underneath, only to find very little water actually stays in the bowl.
Alongside the outsourcing of thought comes the outsourcing of responsibility. Diverging thoughts and actions gradually lose their sense of shame over both process and outcome.
We drift into a state of munyeom-muchi—thoughtless and shameless.
If this is the ethos of the new era, should we conform to it or resist it? There are no physical steam looms left to smash, and it has already seeped into the very capillaries of our lives, so society will likely end up conforming. I have no particular desire to join the resistance.
And yet I find myself caught in the contradiction of not wanting to meekly conform while still wanting to wield its power.
Between conformity and resistance, there must always be a wiser path beyond the two extremes.
That will be the task for all of us in this new era.
To look on the bright side, as inefficiencies in the world are ironed out, it has become possible to focus on more essential things. It has given me a chance to look closely at what exactly it was that I liked and enjoyed under the keyword of “development.”
Simply typing code in itself is obviously not an overwhelmingly thrilling act. Looking back now and mulling over what gave me that sense of efficacy, I suspect it was because the tempo of human thought matched the physical speed of typing code by hand—and the fun lay in navigating the various thoughts that arose in between.
While fingers were busy typing, multiple lines of thought ran in parallel, dynamically updating the very purpose of what was being typed; perhaps it was that ongoing loop that was so enjoyable.
Even in the accelerated tempo of this new era, there will be new kinds of pauses, though it is admittedly hard to keep pace through old inertia alone. Work comes in many forms, and for tedious chores I could not be more grateful… yet it still makes me nostalgic for the past (or have I just romanticized it?).
Lately, I have been going for morning swims. In swimming, both catching the right timing to breathe and actually taking a proper breath at that moment have to align; if anything is off, you quickly run out of breath and sink. Get water up your nose, and you are in for a miserable time.
In a way, rather than having fallen into a state of thoughtlessness and shamelessness, maybe I have simply been thrown into unfamiliar waters. There are no shortcuts; the only way forward is to train hard and keep practicing.
Everyone will have their own methods and training routines, and the experiences we accumulate will grow. What each individual does on their own is meaningful, of course, but isn’t the unique value of being human found in sharing and building upon those experiences together?
Explainable World
We endlessly try to interpret the world. Is it primal curiosity, or a kind of original sin? Beyond mere understanding, we strive to build an explainable world. The ambiguous and the unknown are like the deep abyss of the ocean—objects of fear that simultaneously provoke something within us.
The fundamentals seem largely figured out already. At the physical layer, we know the forces and materials that constitute us, resting upon a foundation of logically independent mathematics.
Yet the dozens of layers stacked on top of that are so highly complex that we still haven’t untangled the relationships between them.
Look at the oak leaf in front of me. Suppose we examine the molecules and constituent materials that make up this leaf. That only gives us a fragment of a snapshot of the leaf; it does not explain the overall operation and ecology of the machine we call an oak tree.
Even the connectome of a fruit fly is, so far, only a snapshot of an individual at a specific point in time; it does not explain what weights it holds, how it is organically wired to physical sensory points, or how it grows.
Isn’t software development the same?
Looking across various projects, even when they sit on the same framework and use the same language, that alone does not represent the ecology and rules of a given project.
Can you understand a project from its source code alone? Just like the oak tree, you have to know the universe and environment in which it exists—how it is deployed, what layers it passes through… It is almost like dealing with a living organism.
Given enough time and intelligence, unlike the oak tree, a human can understand the ecology and behavior of a software project. Having someone who has figured it all out is a huge asset, yet their absence becomes an equally huge risk. Assuming the proportion of what is “well-known” stays constant, a project probabilistically sits in ignorance. Because we fear that abyss, we want to keep it in an explainable state.
LLMs overcome a great deal of that probability. By drastically increasing the pool of “well-known” participants, they improve the odds. Even so, I don’t feel that the proportion of pure participants increases—since the addition comes from the outside.
In understanding the world, if I view myself together with the tools surrounding me as a single logical group, whatever relationship we have with these tools and however we use them in our daily lives, work, and research now and in the future, LLMs help lower the probability of that group being left in ignorance. As participants increase or individual members become more active, those odds improve—yet the better a specific member’s capability becomes, the more dependent the group inevitably grows, and the cognitive expansion of the group and the improvement of individual cognition do not scale proportionally across every element. (Are we in the middle of fading out?)
Coming back to the oak tree: we did not make it. It is an object existing in the same field as us. Software projects, however, were written as collections of explainable things under our cognition. When our cognitive capabilities change, they can be rewritten at any time. Now that everyone works alongside AI, the actual participants are no longer humans as before, but augmented humans—which I think is equivalent to the birth of a new species. Further in the future, of course, we will likely be replaced altogether.