Just thinking... What's the point of high-level programming languages in the future?

This though have been on my mind for a while ever since ChatGPT and other generative AI tools started to gain traction. As of 2025, I see two type of developers, those who use AI tools to assist them in their work and those who "say they don't use it" but actually do.

I remember that first time I was taught about low-level and high-level programming languages at a college class. I was told that high-level languages are easier to read and write, while low-level languages are closer to machine code. That very moment a though came to my mind, "OK so, high-level programming languages are just a product of human incompetence, in the sense that their existance was in a huge part justified by the fact that humans are not capable of writing machine code directly."

I didn't even knew yet how to code when I was taught about this, but ever since then, I couldn't help but see high-level programming as an extra-layer, added for convenience, that wouldn't even exist in an ideal world.

That's why, today, as I look at the current landscape of software development, developers habits and how fast AI assistants are becoming more independent, I can't help but remember that day, years ago, and ask myself: What’s the point of high-level programming languages in the future?

I have shared my concerns with other colleagues and they found my doubts a bit catastrophic, but this is one of those things were I don't want to be right, but I can't help but think I might be right, I might live to see high-level languages become obsolete.

I mean, AI will eventually write 99% of the code for us, the whole point of high-level languages is so that humans talk to machines, but if the future is the machine talking to the machine, then what’s the point of high-level languages? I see it as a middle-step that will eventually become obsolete.

I googled (and asked some LMs to make some research, because why not) a lot about it for the past two years to see if someone else is thinking the same way I am. Surprisingly, I may be indeed a bit catastrophic, there's no so much info about it out there. I mean no one is really thinking about this possibility.

Therefore, I wanted to compilate my own data, which by the way, is a bit old since I didn't compilated the whole thing for this post that nobody is going to read, I compilated it for myself out of curiosity, few months ago.

Please note the AI-generated titles for each section, I thought they were funny and I wanted to keep them.

The Data: AI’s Growing Role in Code Generation

A 2023 GitHub report revealed that developers using Copilot accept 30% of its suggestions and code 55% faster on average. Meanwhile, a McKinsey study estimates that AI could automate 50-70% of coding tasks by 2030. These tools excel at repetitive patterns: generating boilerplate code, debugging, and even translating legacy COBOL to Java.

Yet, AI’s current capabilities are bounded. A Stanford analysis found that LLMs like GPT-4 still struggle with complex logic, contextual understanding, and security-critical code. For example, when tasked with building a secure authentication system, AI-generated code often includes vulnerabilities like SQL injection flaws or improper encryption.

Some conclusions, If high-level programming languages dissapear, might take a while

High-level languages (Python, JavaScript, etc.) emerged to solve two problems: developer productivity and hardware abstraction. AI doesn’t eliminate these needs—it depends on them. Consider:

Legacy systems: 80% of enterprise codebases still rely on languages like Java or C#. Rewriting them in “AI-native” formats is neither practical nor cost-effective.
Collaboration: Code must be maintained by teams. As OpenAI’s Andrej Karpathy noted, “AI-generated code is like a gifted student’s homework—brilliant but inscrutable.” Human-readable syntax remains critical for collaboration.
Precision: High-level languages enforce structure and best practices. AI, by contrast, prioritizes “working code” over “good code,” often ignoring scalability or readability. By the time of writing, this still the main painpoint in my opinion. If you're a developer you know exactly what I mean.

The Hybrid Future: Coding can become a weird type of job

Yeah, this title wasn't generated by AI.

Industry big fishes like Microsoft and Google are betting on a symbiotic future. Their vision? AI handles the “what,” developers handle the “why.” For instance:

Natural language to code: Developers describe a feature (“Sort users by login frequency”), and AI drafts the implementation. But the developer still refines it for efficiency.
AI-powered toolchains: Linters, testing frameworks, and CI/CD pipelines increasingly integrate AI to auto-correct errors or optimize resource usage.
Domain-specific languages (DSLs): New languages tailored for AI collaboration could emerge, blending natural language prompts with traditional syntax. For example, a DSL for data science might allow users to write “plot this data” instead of “create a matplotlib plot.” I have been thinking about this for a while, and I think this is the future of programming languages. I mean, it makes sense, right? This could be the middle-step before high-level programming languages cease to exist. The pre-death phase.

The Risks: Over-Reliance and Skill Erosion

Not all trends are positive. A 2022 Stack Overflow survey found that 62% of developers fear AI could erode foundational coding skills, akin to how calculators impacted mental arithmetic. Over-dependence on AI risks creating a generation of engineers who understand prompts but not principles—a liability when systems fail or requirements evolve. In other words, engineers who don't know shit about programming.

Conclusion: Nobody knows shit

Even if high-level languages persist, their role will evolve. Developers may spend less time writing loops and more time designing systems, auditing AI output, and solving higher-order problems. As MIT’s CSAIL director Daniela Rus puts it: “AI won’t replace programmers—it will replace programmers who don’t use AI.”

The future belongs to those who wield both code and context. Syntax may fade in prominence, but the art of structuring logic—whether in Rust or via a prompt—will remain irreplaceably human.

— Written with AI assistance for research, but argued and edited by a developer who still cares about semicolons.

Hope you enjoyed this stuff,