When I moved into the

I've seen claims that AI will replace skilled workers, not newcomers. It's a bit of a wild story. No, right now I've managed to replace 20 juniors with AI, but I still need "seniors". Why is that?

When I moved into the Seoul National University Research Park in 2015, there was a company called Sua Lab next to it. This was a company that did machine vision AI. For example, it was a company that assembled a product and made a solution that could recognize defects without people passing in front of a camera. For your information, it is an exciting and wonderful company at a good price.

This is how the early machine learning revolution in the past worked to replace the "skilled worker" who can recognize the defect at a glance. But this is more than a decade ago.

The original article claiming that AI is not a threat to new employees is based on the fact that new employees have become deskilled, not that new employees have disappeared due to the advent of office automation. I agree with that basis, and rather explain why the story implies a crisis for new employees.

In 2022, ChatGPT (LLM), which has grown the size of the model on a previously unimaginable scale, appeared. From this point on, computers can perform undefined and unskilled tasks. In other words, the advent of LLM has created a "de-skilled" of existing computer agents. These de-skilled computer agents have begun to overlap with the positions of "de-skilled newcomers," as they claim.

To sum up, it's a long time ago that skilled workers started to be replaced, and new recruits are now starting to be replaced. So the argument "replacing skilled workers" becomes a belated argument without nutrition, and "new recruits are not replaced" becomes a wrong argument. That's why the entire text became a wrong one, which is not easy to refute.

If so, you may wonder why "senior" is needed when skilled workers have already begun to be replaced. It's the same that seniors, juniors, and AI are more expensive. In context, when it comes to the need for a senior in the AI era, this senior is not a skilled senior, but a senior as a designer or manager.

The work of modern companies follows the structure in which seniors design big pictures and strategies, juniors upload frames, get approval, seniors approve them again, and juniors execute them. Making frames here? The Generative AI does it like a genius. Execution? When combined with Agent, it ends. However, designing, approving, and taking responsibility for big pictures and strategies is difficult for computers to replace. That's why juniors are replaced faster than seniors.

I hope that people who have not personally used AI properly and have not used a single line of agent code will refrain from writing grand discourses by even using expressions that are ignored by others. I think I said the same thing before, but this country seems to have more AI influencer than AI researchers and developers.