If AI Is Now Writing Our Code, Do We Still Need Engineers?

In a recent episode of Y Combinator’s Lightcone podcast, the hosts shared a remarkable statistic. In the current YC cohort, as much as 95 percent of the actual character code written for platforms is generated by AI. That means most of the code shipping in some of the world’s most promising startups is being authored not by human hands, but by machines.

This emerging reality, referred to as "Vibe Coding" in the podcast, marks a major shift in how software is built. Founders and teams are now describing what they want, prompting the AI to generate it, and then refining the results. It is fast, flexible, and already changing the role of the engineer.

That raises a larger question. If AI is writing the code, where does that leave the engineers who used to write it?

The Changing Role of Engineers

The podcast also highlighted another important trend. As AI takes on more of the raw code generation, front-end engineers are increasingly moving closer to product management. Their understanding of user experience and rapid iteration cycles makes them a natural fit for product definition. Back-end engineers, on the other hand, are focusing more on cloud infrastructure and system configuration. They are spending their time on architecture, deployment, and ensuring performance and scalability.

This suggests that engineering as a discipline is evolving rather than disappearing. We are seeing a shift from manual code writing to higher-order responsibilities. Engineers are becoming architects, integrators, and decision-makers. The center of gravity in software development is moving toward strategy and systems thinking.

Cheating the Basics, Revisited

Back in 2009, I wrote a blog post titled “Cheating the Basics” The core idea was that both great and poor athletes cheat the basics, but only the great ones do so successfully. The difference is that they first master the fundamentals. They know when to break the rules because they understand them deeply.

This same principle now applies to AI-based coding.

Yes, AI can write vast amounts of code. But when that code fails to compile, when edge cases cause unexpected behavior, or when performance suffers, you still need someone who understands how everything fits together. AI can generate solutions quickly, but it does not understand intent, tradeoffs, or long-term design implications. That is the job of a skilled engineer.

Great engineers will use AI to extend their reach. They will iterate faster, explore more possibilities, and build better products. Engineers without a solid foundation will still be able to produce more than they could in the past, but they will eventually hit a wall. And when that happens, they will not know how to diagnose the problem or correct the course.

Experience Still Matters

The rise of AI does not reduce the need for engineering expertise. If anything, it raises the bar. As software becomes more interconnected and AI-driven, the systems we build are becoming more complex, not less. When things go wrong, you need someone who can read between the lines of auto-generated code, troubleshoot cloud infrastructure, and understand how all the pieces interact.

In other words, AI is writing the code, but engineers are still essential. Their role is shifting from typing syntax to understanding architecture, design principles, and user needs. The most valuable engineers will be those who have mastered the fundamentals and now know how to apply them in an AI-enhanced environment.

We are entering a new era of engineering. Those who have invested in understanding the discipline will thrive. Those who rely on AI without building that foundation may appear productive in the short term, but will struggle when challenges arise.

Just like in sports or leadership, those who have earned the right to skip steps are the ones who understand what those steps mean. And they are the ones who will continue to lead in this new world.

How do you see AI impacting the future of software engineers? Add your comments below.