AI is often perceived as a replacement for humans, but in reality, it is far from that. It’s not about the technology itself, but about how it’s perceived. Humans have the ability to practice, build patterns, and improve in a niche over time. AI still does not “learn on the job.” At present, one of the most widespread use cases of AI is software development, yet even here, the tools do not improve at runtime. They require constant orchestration, critical verification, and validation of outputs. These tasks may sound simple but demand deep knowledge and understanding of the systems involved.
Since the Industrial Revolution, people have predicted that human labour would become obsolete. Machines were expected to replace us, and indeed, they made many processes more efficient and production easier. Yet here we are in 2025, and I am using a machine to write this article—human labour is far from gone. The reason is simple: humans adapt, create new economies, and thrive within them. This tradition of evolution is still difficult for AI to replicate.
For example, I would never hire an AI designer in place of a human one. A human designer understands my taste, my product, and can actively learn from and question my choices. LLMs can mimic this to some extent, but the assumption that they truly “get it” is still hard to accept.
This is why I believe AI will go through a Dunning–Kruger–style cycle. Initially, many will overestimate its capabilities and use it as a human replacement. Over time, they will realise these systems are not being challenged in radically creative ways—the kind of challenge only humans can bring. This will increase demand for genuine insight and experienced professionals. The road to such experience may still require years of deep, focused learning, but tools will exist to support that journey. And while we live in a capitalist world, this reality will continue to shape how AI fits into human progress. Although there will be huge productivity jump in systems. I think it would be in order of magniute of cloud jump and not computer jump. So ideally how everything moved to cloud brought in productivity improvement, it was an era when you heard the words like (social local global cloud data science) for the first time. It was also the time when the phrase "data is the new oil" became extremely popular. This phase started in 2005, and followed till 2015. The AI era, which will enable people to orchestrate their jobs with bunch of prompts especially any kind of a soft job, which does sound scary but will eventually lead to new things quickly which weren't possible with AI before in not just software industry but in other industries as well.
The quote (which I know most people in tech are tired of hearing) of "If I had asked people what they wanted, they would have said faster horses", is so much true today.
Posts
