Until now, software developers were in a way similar to tailors who sewed clothes one stitch at a time. Some automation tools (such as code autocompletion and static analysis tools) helped with coding itself or debugging, but fundamentally, software development was a manual process that humans had to perform from start to finish.
This was true for individuals and companies alike. Large software development companies simply employed a lot of software developers. Just as factories existed before the invention of sewing machines/spinning machines/looms, where large numbers of tailors were gathered to produce fabrics and clothes by hand, software companies operated in a similar way. They employed a large number of artisans who used their craftsmanship to create products (clothing, software).
AI agents have drastically changed the software development process. While vibe coding opened the door for non-experts to enter the world of software development, for professional software developers, AI agents brought about a significant increase in productivity. It was like when tailors who sewed clothes by hand first encountered sewing machines.
While it's unclear whether it's AI washing or not, on the surface, almost all software companies are laying off employees en masse due to the surplus of manpower caused by AI. There are rumors that senior developers who were laid off are being rehired, but the undeniable fact is that software developers currently have mixed feelings about AI. They can't work without AI, and they also worry that AI might take their jobs. I feel the same way.
So, I desperately think about what our future holds. It's not just me. These days, when I chat with colleagues at work, this is the only topic we discuss. Will our livelihoods be safe in the face of the powerful wave of AI and its ever-improving capabilities? Will I be able to make a living until my child becomes independent? Why do I have to worry about making a living now when retirement is still years away? My senior colleagues who are close to retirement are the most enviable. And so on.
Since we can't predict the future, we can only speculate by looking back at the past and extrapolating from it. That's why I mentioned tailors in a hurry. At this point in time, AI agents are certainly not perfect. Skilled human engineers' control and insight are still needed. Nothing progresses without human intervention at this stage. This is a clear fact. However, the problem lies in the speed of AI development. This may be similar to what tailors who sewed clothes by hand felt 200 years ago when sewing machines/spinning machines/looms (hereinafter referred to as sewing machines) were invented and spread.
....
Let's imagine now, going back 200 years, what they would have been like. Clothes were difficult to make, and especially good quality, well-designed clothes could be sold for a high price. As a result, my value was also high. Wealthy people who could afford to buy clothes were demanding customers who paid attention to the quality of every stitch. Putting design aside, there weren't many people who could sew as well as I could. It took me decades to perfect my sewing skills. Then one day, something called a sewing machine appeared. This machine effortlessly sewed those difficult stitches and did so with perfect accuracy down to the last millimeter. My decades of experience became meaningless.
That's not to say that a sewing machine is something anyone can just pick up and use. Some practice was needed. But the time it took to reach a certain level was incredibly short. The sewing quality of a tailor with ten years of experience was comparable to that of someone who had practiced using a sewing machine for three months. We tailors, including myself, could no longer compete based on our sewing skills alone. Moreover, I also used a sewing machine. While I still possessed the skill to sew better than a sewing machine, the demand for such high-quality sewing was lower. The quality of clothes produced by a sewing machine was sufficient for most people.
As a result, the time it took to make clothes decreased. Naturally, clothing production increased. Clothing prices dropped. A colleague I knew bought a large number of sewing machines and trained people to use them, setting up a factory that could produce a lot of clothes quickly with sewing machines. He used to sell ten 100-won clothes a month, but now he sells 5,000 10-won clothes. He's making 50 times more money. Of course, you have to calculate the factory workers' wages and maintenance costs to determine the net profit, but even then, it's said that he's making a lot more money than before. He came to me and asked if I would work at his factory. He said they needed skilled workers to train employees, inspect and manage the quality of the products and the production process.
....
As I was imagining, I naturally ended up thinking about something similar to what happened during the Industrial Revolution. That's right. When the cost of producing a commodity drops sharply, the price of that commodity falls. As a result, consumption of that commodity increases, and ultimately, the market size for that commodity grows, leading to an increase in demand for workers who produce that commodity. This is the economic cycle we have experienced so far. Will this past rule apply again to software development? I personally hope so. I still have a long way to go before retirement.
(Of course, you can imagine other scenarios just as bleak, but I don't want to write them down and share those thoughts.)
