One of the things that I used to like about programming by hand was to find a way to catch all requirements in a balanced and complete set of statements that left no room for error or misinterpretation. With the right level of abstraction, yet still be understandable enough to be reasoned about without great effort.
To me this is the essence of the software engineering discipline: connecting the dots, mapping the exceptions and edge cases that are implicit in the requirements and transforming them into an executable application that meets the explicit and implicit expectations of the client.
But all that was still a translation of natural language into code. With AI we hardly need to do that translation by hand anymore. That doesn't mean the engineering discipline disappears. On the contrary! It moves it into the realm of natural language, removing the burden of the translation.
Just like not everyone is a good salesperson, not everyone is a good CEO, project manager, designer or engineer. Our qualities as engineer are also pointing out the obvious and subtle errors and contradictions in specifications before the application is built. And that doesn't change.
To me this is the essence of the software engineering discipline: connecting the dots, mapping the exceptions and edge cases that are implicit in the requirements and transforming them into an executable application that meets the explicit and implicit expectations of the client.
But all that was still a translation of natural language into code. With AI we hardly need to do that translation by hand anymore. That doesn't mean the engineering discipline disappears. On the contrary! It moves it into the realm of natural language, removing the burden of the translation.
Just like not everyone is a good salesperson, not everyone is a good CEO, project manager, designer or engineer. Our qualities as engineer are also pointing out the obvious and subtle errors and contradictions in specifications before the application is built. And that doesn't change.