Skip to content

#300: Learning Python in the Age of AI?

As AI tools grow in popularity and AI agents for code generation become more common, the question naturally arises: Should we even bother learning Python anymore? Or is it a waste of time when the machine can generate the application for us?

The more I think about it, the harder it gets to find a satisfying answer. Many of the cherished reasons of the past will disappear. Others no longer matter that much when we can change large parts of code in an instant. Let us explore the main points I came up with that we should consider.

The promise vs. the reality of AI code generation

If all the hype were true, we could stop learning programming entirely. We would describe our requirements, and the AI would generate perfect, efficient code every time. Why invest years into mastering a language when a prompt could do it for us?

But in practice, this is not the world we live in. AI agents like Junie can generate useful solutions and save hours of work, but they also drift, fabricate packages, create security problems, and make mistakes. When that happens, someone still needs to know:

  • what went wrong,
  • why it went wrong, and
  • how to fix it without starting over from scratch.

That "someone" is us. Which means that understanding Python - at least well enough to debug and repair AI output - is still essential. That is true for every language, but here in this Python blog we focus on Python.

Skills in the age of AI

If AI is changing the landscape, what skills matter most now?

  • Skills that stay in demand: The ability to define problems clearly. What do we want to achieve, and why? Requirements gathering, systems thinking, and separating "needs" from "wants" will remain critical - because AI can not clarify our goals for us.
  • Skills losing importance: Memorizing syntax and library details. AI can already provide this on demand. With retrieval-augmented systems, it can even explain stack traces and fix common errors. This reduces the need to know a library by heart.
  • Skills often neglected but crucial: Software engineering principles like testing, security, and maintainability. These matter even more in the AI era, but they are as usually ignored. The problem is that failures in these areas show up with a delay - often only when it is too late.

Python continues to be an outstanding tool for developing proficiency in these categories: it is accessible for learning foundational concepts, sufficiently flexible for addressing complex problems, and widely adopted across testing and engineering domains.

The economics & sustainability of AI tools

Another overlooked factor is cost. Today’s AI services look cheap, but mostly because they are subsidized by venture capital. The real cost of running these models is much higher. If those subsidies vanish, access to the tools could get 10x, 20x, or 50x more expensive.

And then there is sustainability. AI is in a hype cycle. Like every bubble before, it will eventually burst. The only question is when that is going to happen, not if. Many tools will vanish, but some of the genuinely useful solutions will remain.

What does this mean for learners? It means relying solely on AI is risky. If prices spike or services vanish, the ability to write and understand Python yourself becomes your safety net.

How to approach learning Python today

Given all those considerations, the way to learn Python is not about memorizing every detail - it is about cultivating the skills that AI can not replace. Python becomes the medium for developing those skills:

  • Practice problem definition: Use Python projects to translate vague ideas into concrete requirements. For example, instead of "I want to analyse sales", define precisely what metrics you need, how the data is structured, and what outputs matter. AI can generate code, but you need clarity to guide it.
  • Focus on systems thinking: Do not just write scripts - design small systems. Even a Python data pipeline or an automation task is an opportunity to think about inputs, outputs, dependencies, and failure points. This develops the mental models that scale far beyond Python.
  • Lean on AI for syntax, but not for structure: Let AI fill in the boilerplate but take responsibility for shaping the architecture. This means understanding how modules fit together, when to use functions vs. classes, and how to organize code so it remains understandable.
  • Strengthen engineering discipline: Treat every project as if it were going to production. Write tests, handle errors, and think about maintainability. Python’s ecosystem makes this easy to practice - and these are exactly the skills AI-generated code often lacks.
  • Build what matters to you: Projects tied to your interests or goals will give you motivation and context. Whether that is automating tasks, analysing data, or experimenting with AI libraries, the skills you gain will transfer directly to future challenges.

In other words: do not learn Python to compete with AI at producing lines of code. Learn it as a way to think clearly, design responsibly, and collaborate effectively with the tools of the future.

Conclusion: Should We Still Learn Python?

The short answer: Yes. But not in the same way we did before.

We do not learn Python to outcompete AI at producing code. We learn it to understand problems, to collaborate effectively (also with AI), and to build a solid foundation in software thinking.

Python can not only help us cultivating the more general skills but also help us with building our own AI systems. That makes Python for me the ideal language in these uncertain times. If AI usage increases and the bubble does not burst for a few years, we are covered. And if it burst, the automation capabilities of Python will be especially handy to replace the no longer available tools.