14. When Code Gets Cheap, Where Does a Developer's Value Go

Lately I read Andrew Ng's new post on X, AI Engineering Skills Map: Shaping the build, and it made me, an engineer of more than ten years, anxious about my career.

The core is one sentence: a strong software engineer in the AI era has to move past implementing requirements someone else defined and start taking part in defining what gets built, how, and how well.

Ng condenses the role into "shape the build."

The explosion of AI coding tools pushed the cost of generating code toward zero. Driving the whole process becomes the main way to stand out. With these tools an engineer can validate an idea, build a prototype, ship an MVP, get feedback, and change the product without waiting for each other role to finish first.

I think the position of an engineer in the AI era fits this formula, and it will probably keep changing:

AI Engineer = Engineering × Product Sense × Communication × Agency

The post also implies a bigger claim: the value of an engineer will shift from producing code to judging what is worth producing.

Once the boundaries between product, design, and engineering blur and then break, nobody's work stays limited by their own skill boundary. Judgment, trade-offs, user understanding, ownership, and a sense of direction get more valuable instead.

Engineers sit closest to the code, so we need to push the build loop forward and own the pace of iteration on the project and the product. That will be the first core barrier in the AI era. I am still working toward becoming a well-rounded engineer, and a job title that looks about to become history makes the future hard to feel calm about.

Plus:

The DeepLearning.AI ad at the end of that post came out of nowhere. At least the piece itself reminds me to keep learning.