The relentless advance of artificial intelligence is not only revolutionizing entire sectors, but it is also transforming the software engineering profession itself. Mike Krieger, co-founder of Instagram and current product head at Anthropic, has been clear in a recent interview on the 20VC podcast: within three years, engineers will spend more time reviewing code generated by AI than writing it themselves.
From Programmers to Quality Supervisors
According to Krieger, the role of the engineer will evolve into a more strategic and high-level profile, focused on validating, adjusting, and overseeing the code created by advanced language models like Claude, from his own company. “It’s about finding the right ideas, properly designing user interaction, and verifying everything at scale—something that only a human team can deeply evaluate,” he explained.
This shift does not imply the disappearance of the engineer role but rather a redefinition of its skills. AI can handle repetitive tasks and the mass generation of code, but supervision, deep understanding of objectives, and context remain irreplaceable tasks for now.
Is AI a Threat or an Ally?
The debate over whether artificial intelligence will replace developers remains open. Dario Amodei, CEO of Anthropic, has gone a step further, claiming that in the next 3 to 6 months, AI could write up to 90% of the code, and in less than a year, virtually all of it. However, Amodei himself acknowledges the necessity of engineers, as evidenced by the strong hiring at his company.
Mark Zuckerberg, CEO of Meta, takes a more cautious approach and predicts that by 2025, AI will be capable of writing code at the level of a mid-level engineer, but always with humans overseeing the process.
The Current Reality: AI as a Productivity Tool
The present already points toward that future. The 2024 developer survey from Stack Overflow showed that 82% of programmers use AI tools to write code. Similarly, JetBrains estimates this figure at 69%.
GitHub highlights an increase of up to 55% in productivity thanks to code assistants, and most revealing: nearly 90% of developers in the United States and over 80% in India believe these tools improve the quality of the software they produce.
Krieger sums it up clearly: AI is not going to take away jobs from software engineers; it is going to transform their work. The key will not just be programming, but being able to manage, review, and guide increasingly autonomous systems. In other words, humans will remain essential in the process, even as their role drifts further away from the keyboard and closer to leadership and strategic oversight.
References: Error500, Hipertextual, and El chapuzas informático.