Pseudocode is the New Code

Yesterday I went to a workshop put on by the Carnegie Mellon Science and Technology Library hosted at CoLab 18 here in Pittsburgh. The topic was how to use AI and AI agents to assist in programming.

First, a quick definition. Pseudocode is a way of describing a program’s logic in plain language without needing exact syntax. It reads more like a mix of English and programming, focusing on the structure of an idea rather than the exact characters a compiler needs. Think of it as sketching a blueprint before drafting architectural blueprints.

One of my main takeaways, which has also been borne out in my own recent experience, is that pseudocode is becoming the new code. Not entirely, and many software engineers might argue with me, but at this point what is essentially more important than the ability to produce exact syntax is the ability to structure your ideas logically. In other words, to be able to program verbally or by typing out pseudocode.

Of course, code is based around Boolean algebra, which is a way of formalizing logical statements. That is why we often refer to programming as writing logic. But now we have very advanced natural language processing. The original purpose of code was to create a layer of abstraction so experts could give commands to a computer, which would then be translated into the ones and zeros of machine code. Now we have another layer of abstraction, where large language models can produce the machine-readable syntax themselves.

The catch is that these models are not very good at understanding context. Without a human in the loop, they struggle to produce coherent logic. But if you can state what you want done in clear pseudocode, the AI is remarkably effective at translating that into functioning code in whatever language you need.

So what does this mean? In practice, it means that learning to think and communicate in pseudocode is now a critical programming skill. The human role is shifting toward clarity of thought and logical design, while the machine handles the syntax. Being able to sketch your ideas in pseudocode gives you leverage: the computer does the heavy lifting, but you remain in charge of the reasoning.

Copyright © 2025 Ryan Badertscher. All rights reserved.